ICINCO 2025 Abstracts

Area 1 - Industrial Informatics

Full Papers

Paper Nr:	25
Title:	An Efficient and Scalable Hyperdimensional Computing Framework for Anomaly Classification in Industrial Systems
Authors:	Víctor Ortega, Soledad Escolar, Fernando Rincón, Jesús Barba and Julián Caba
Abstract:	This paper presents a hyperdimensional computing (HDC)-based framework for anomaly classification, designed to meet the specific demands of industrial systems. Inspired by cognitive processes, HDC employs high-dimensional representations to enable robust, low-complexity, and hardware-efficient computation. The proposed framework encompasses the entire processing pipeline, from data encoding to anomaly classification, and is optimized for efficient execution on both conventional computing platforms and resource-constrained devices. To assess its effectiveness, we conduct a case study based on a real-world scenario involving 118 emergency lighting devices that collect and transmit operational data to a central sink capable of detecting anomalous behavior. Experimental results demonstrate that the proposed approach achieves high classification accuracy and confirm its suitability for deployment in integrated industrial systems with limited computational resources.
Download

Paper Nr:	91
Title:	Semi-Supervised Anomaly Detection in Directed Energy Deposition Using Thermal Images
Authors:	Ufuk Ismail Ozdek, Yigit Kaan Tonkaz, Shawqi Mohammed Farea and Mustafa Unel
Abstract:	Directed Energy Deposition (DED) is a crucial additive manufacturing process used in aerospace and healthcare applications, among others. However, ensuring defect-free production remains a challenge due to the difficulty in detecting defect-related anomalies in real-time. In this study, we address the problem of defect detection in DED processes through thermal images of melt pools. As an anomaly detection problem, we adopt a semi-supervised approach based on One-Class Support Vector Machine (OCSVM) and Isolation Forest (iForest). We analyze the performance of these models across different feature sets. Additionally, this semi-supervised approach is compared against an unsupervised approach utilizing the same learning algorithms. The results indicate the superiority of the semi-supervised approach for both algorithms. Yet, iForest outperforms OCSVM with an accuracy of 95% and an F1-score of 0.88, demonstrating its robustness in distinguishing defective from non-defective instances. This work provides valuable insights into the applicability of semi-supervised machine learning techniques for real-time defect detection in DED processes. By leveraging thermal imaging data and feature-based anomaly detection models, our findings contribute to the development of efficient, non-destructive quality control mechanisms for additive manufacturing processes.
Download

Paper Nr:	125
Title:	Real-Time Automated Visual Inspection of Decorative Wood Panels for Zero Defects Manufacturing
Authors:	Beatriz Coutinho, Tomás Martins, Eliseu Pereira and Gil Gonçalves
Abstract:	In the wood panel manufacturing industry, maintaining high product quality is critical to ensure customer satisfaction and minimize resource waste. Manual quality inspection methods are often inconsistent, increasing the risk of defective panels reaching the market. This paper introduces an automated visual inspection system for decorative wood panels, aligned with the Detection strategy of the Zero Defects Manufacturing (ZDM) framework. Designed for real-time deployment on an NVIDIA Jetson Nano, the system inspects panels independently without disrupting the production line and visually highlights detected defects for operator review. Two implementation approaches were explored and compared: a traditional computer vision pipeline and a deep learning-based solution. Due to the limited availability of real-world defect images, a synthetic dataset was created using patch blending, tiling, and diverse augmentations to improve the model’s generalization across spatial variations. Experimental evaluation with static images and live video showed that while traditional methods achieve moderate detection accuracy, they fail under varying lighting and camera angles. In contrast, the YOLO-based approach delivered robust segmentation and superior defect detection, even under challenging conditions. These results highlight the system’s potential to assist operators during manual inspections and contribute to practical advances to achieve ZDM.
Download

Paper Nr:	138
Title:	Improving Industrial Interoperability and Scalability Through OPC-UA and Smart Object-Based Architectures
Authors:	Guilherme Coelho, Liliana Antão, Beatriz Coutinho, Gil Gonçalves, António Augusto and Miguel Moura
Abstract:	This paper presents the development and implementation of a remote monitoring and control system for industrial machines, aligned with the principles of Industry 4.0. The proposed solution builds under the Advanced4i work package of the PRODUTECH R3 initiative, addressing key limitations in industrial digitalization scalability, responsiveness, and usability. A redesigned architecture is introduced, build upon a previous data model and architecture by Neto et al., featuring advanced communication protocols, a refactored LabVIEW-based interface, and a middleware layer to enhance data flow and synchronization. A structured data model and an optimized graphical user interface further enable real-time monitoring and remote configuration. The system was validated in a real-world deployment at IDEPA, a real manufacturing company a leader in the labels, tapes, and textile accessories markets , integrating over 30 sensors, and subsequently scaled to support up to 85 sensors with minimal data loss and high responsiveness under operational stress. The results demonstrate capability of maintaining monitoring performance and scalability, offering a practical roadmap for the deployment of interoperable and modular Industry 4.0 solutions in manufacturing environments.
Download

Short Papers

Paper Nr:	48
Title:	Noise-Robust Speech Transcription with Quantized Language Model Correction for Industrial Settings
Authors:	Marco Murgia, Marco Fontana, Alberto Pes, Diego Reforgiato Recupero, Giuseppe Scarpi and Leonardo Daniele Scintilla
Abstract:	In this paper, we propose a robust and computationally efficient pipeline for transcribing speech in noisy environments, such as workshops and industrial settings. The pipeline is designed to operate offline, making it suitable for resource-constrained scenarios. It begins with a noise filtering module that preprocesses audio recordings to suppress background noise and enhance speech clarity. The filtered audio is then passed to an Automatic Speech Recognition (ASR) model, which generates initial transcription outputs. Given the potential for transcription errors in challenging acoustic conditions, we incorporate a quantized Small Language Model (SLM) trained on an ontology of defects related to the industrial environment to post-process and correct these errors. The quantization of the SLM significantly reduces its computational footprint while maintaining correction accuracy, enabling the pipeline to function effectively on low-resource devices. Experimental evaluations demonstrate the effectiveness of the proposed approach in improving transcription quality in noisy conditions, highlighting its practicality for offline and resource-limited applications. In fact, preliminary validation on a synthetic dataset of 200 sentences in Italian and English showed a consistent F1 score of 87.04% for SNR as challenging as -5 dBW (Decibels Watt) in Italian sentences and 91.25% in English sentences, with the least computationally expensive version of Whisper (Whisper Tiny) and the SLM correction.
Download

Paper Nr:	51
Title:	Fault Diagnosis of Industrial Robots Using a Digital Twin and GRU-Based Deep Learning
Authors:	Ilhem Ben Hnaien, Eric Gascard, Zineb Simeu-Abazi and Hedi Dhouibi
Abstract:	This paper proposes a fault diagnosis method for industrial robots based on the combination of a digital twin and a GRU-based deep learning model. A high-fidelity digital replica of the 6-DOF Stäubli TX60 robot was developed using MATLAB Simulink and Simscape Multibody to simulate both normal and faulty behaviors. A dedicated fault injection module was used to generate motor blockage scenarios at different time instants, creating a labeled dataset of 49 classes. The time-series data were then used to train a Gated Recurrent Unit (GRU) neural network, which is efficient for modeling temporal patterns. The trained model achieved an accuracy of 87.35%, with strong performance across different fault types. This approach enables reliable, non-invasive, and repeatable fault diagnosis and provides a solid foundation for future work on predictive maintenance and deployment on real robotic platforms.
Download

Paper Nr:	76
Title:	Towards Industry 5.0: AAS/MLOps-Driven Model Maintenance for Data-Centric Production
Authors:	Kiavash Fathi, Marcin Sadurski, Stefan Waskow, Tobias Kleinert and Hans Wernher van de Venn
Abstract:	Despite the advancements brought by digitalization across industries, only a few state-of-the-art data-driven methods successfully transition to production and remain viable. The sheer volume of physical assets in production lines, combined with constantly evolving requirements, makes model deployment and maintenance highly complex. This paper presents a production-ready architecture developed for data-driven digital assets at ABB Schaffhausen AG. The solution integrates MLOps best practices orchestrated via MLRun with the industry-standard metadata modeling system, Asset Administration Shell (AAS). We demonstrate how controlled artifact generation from MLRun facilitates experiment tracking and knowledge sharing while AAS ensures standardization and long-term maintenance. By combining MLOps and AAS, we effectively manage the ever-growing artifacts of data-driven solutions. Additionally, we explore how controlled artifact generation enables role-based MLOps by restricting access to relevant information based on industrial roles. This architecture supports a smooth transition to Industry 5.0.
Download

Paper Nr:	92
Title:	Anomaly Detection in Directed Energy Deposition: A Comparative Study of Supervised and Unsupervised Machine Learning Algorithms
Authors:	Berke Ayyıldızlı, Beyza Balota, Kerem Tatari, Shawqi Mohammed Farea and Mustafa Unel
Abstract:	Directed Energy Deposition (DED) is a promising additive manufacturing technology increasingly utilized in critical industries such as aerospace and biomedical engineering for fabricating complex metal components. However, ensuring the structural integrity of DED-fabricated parts remains a significant challenge due to the emergence of in-process defects. To address this, we propose a comprehensive anomaly detection framework that leverages in-situ thermal imaging of the melt pool for defect identification. Our approach encompasses both supervised and unsupervised machine learning techniques to capture diverse defect patterns and accommodate varying levels of labeled data availability. Supervised methods-including ensemble classifiers and deep neural networks-are employed to learn from annotated thermal data, while unsupervised methods, such as autoencoders and clustering algorithms, are used to detect anomalies in unlabeled scenarios and uncover previously unknown defect patterns. The pipeline incorporates essential preprocessing techniques-such as feature extraction, normalization, and class rebalancing-to enhance model robustness. Experimental evaluations offer a detailed comparison between the supervised classifiers and unsupervised models utilized in this work, emphasizing the predictive performance and practical applicability of each learning paradigm. Notably, the supervised classification-based framework achieved high performance in detecting porosity-related anomalies, with an F1 score of up to 0.88 and accuracy reaching 99%.
Download

Paper Nr:	33
Title:	Safety-Centric Monitoring of Structural Configurations in Outdoor Warehouse Using an UAV
Authors:	Assia Belbachir, Antonio M. Ortiz, Ahmed Nabil Belbachir and Emanuele Ciccia
Abstract:	In industrial warehouse environments, particularly in steel bar manufacturing scenarios, ensuring the structural stability of stacked bars is essential for both worker safety and operational efficiency. This paper presents a novel vision-based framework for automatic safety validation of outdoor storage bays using a dual-resolution implementation of the Segment Anything Model (SAM). The system processes video streams coming from drone (AUV) by combining zero-shot segmentation with geometric reasoning to assess lateral and frontal support conditions in real time. At each frame, SAM is applied at two scales to extract both fine-grained support components and large bulk regions. A morphological proximity rule reclassifies unsupported regions based on contact with multiple smaller support masks. Additionally, a frontal-view analysis computes bar-end centroids and applies a triangle-based inclusion test to determine correct placement. Experimental results on real warehouse videos demonstrate robust safety classification under occlusion and clutter, with interactive frame rates and no need for manual annotation. The proposed framework offers a lightweight, interpretable solution for automated safety monitoring in complex industrial environments.
Download

Paper Nr:	107
Title:	Smart Optimized Scheduling Under Constraints in Industry 5.0 Through Intelligent Computational Methods
Authors:	Cherifa Nakkach, Wiem Abbes and Yvan Picaud
Abstract:	Production scheduling has become an integral component of next-generation industrial systems during the era of Industry 5.0, which emphasizes collaboration between humans and machines, sustainability, and hyper-personalization. To address complex scheduling challenges, this paper presents a smart scheduling framework based on metaheuristic optimization tailored for manufacturing environments incorporating 3D printing technologies. The proposed framework addresses several key objectives, including the optimization of energy consumption, efficient utilization of raw materials, and minimization of total production time. By incorporating metaheuristic algorithms such as Genetic Algorithms, Particle Swarm Optimization, and Ant Colony Optimization, the system demonstrates adaptability to multiple constraints and competing priorities. Experimental evaluations confirm the framework’s effectiveness in enhancing operational efficiency, flexibility, and sustainability, in alignment with the core principles of Industry 5.0.
Download

Paper Nr:	123
Title:	Leveraging Edge and Fog Resources While Complying with EU’s GDPR
Authors:	Matilde Silva, Pedro C. Diniz and Gil Gonçalves
Abstract:	In Industry 4.0 environments, video-based monitoring systems must now reconcile performance demands with the privacy mandates of the European Union’s (EU) General Data Protection Regulation (GDPR). This paper presents a fault-tolerant edge/fog architecture designed to anonymize visual data at the point of capture, minimizing personal data exposure while maintaining low-latency analytics. Built on the IEC 61499 standard, the system uses DINASORE to run Function Blocks directly on edge devices, and T-Sync as an orchestrator that dynamically reallocates tasks in response to topology changes. Empirical evaluations demonstrate that the architecture reliably recovers from node loss and stays within resource limits even on modest hardware. Despite bottlenecks under heavy vision workloads, the results show the viability of deploying GDPR-compliant IIoT pipelines without offloading sensitive data to the cloud.
Download

Paper Nr:	143
Title:	Digital Transformation of the Nuclear Industry: Leveraging Robotics, AI, and Digital Twins for Standardised, Safe, and Efficient Operations
Authors:	Abdenour Benkrid, Omar Zahra, Réka Szőke, Ankur Shukla and István Szőke
Abstract:	The nuclear back-end is experiencing a pivotal digital transformation driven by the integration of robotics, artificial intelligence (AI), and digital twin (DT) technologies. These innovations hold strategic potential to enhance safety, efficiency, and standardisation across decommissioning, waste management, and site remediation. Using the Technical, Economic, Commercial, Organisational, and Political framework (TECOP) and Five Case Models, this paper critically assesses the value, deployment readiness, and integration barriers of these digital tools across technical, organisational, and regulatory domains. Emphasis is placed on robotics in 5D contexts, the nuclearization challenge, and the role of DT and Building Information Modelling (BIM) in scenario planning and compliance. Persistent obstacles, including fragmented certification, cybersecurity vulnerabilities, limited interoperability, and resistance to change, are analysed using data from expert surveys and project experience. Targeted strategies are proposed to address these issues and accelerate technology readiness and regulatory harmonisation. The contributions of EU-funded initiatives such as HARPERS, DORADO, and XS-Ability are highlighted as catalysts for safe and scalable digital innovation. By providing actionable recommendations, this paper supports policymakers, industry leaders, and technology developers in advancing the digital evolution of the nuclear back-end.
Download

Area 2 - Intelligent Control Systems and Optimization

Full Papers

Paper Nr:	21
Title:	Enhancing Pharmaceutical Batch Processes Monitoring with Predictive LSTM-Based Framework
Authors:	Daniele Antonucci, Davide Bonanni, Domenico Palumberi, Luca Consolini and Gianluigi Ferrari
Abstract:	Monitoring industrial processes and understanding deviations is critical in ensuring product quality, process efficiency, and early detection of anomalies. Traditional methods for dimensionality reduction and anomaly detection, such as Principal Component Analysis (PCA) or Partial Least Squares (PLS), often struggle to capture the complex and dynamic nature of batch data. In this study, we propose a novel approach that combines an AutoEncoder (AE), based on Long Short-Term Memory (LSTM) layers, with a rolling threshold for anomaly evaluation. Unlike conventional threshold methods that rely on global statistical parameters, the applied threshold leverages rolling median and rolling Median Absolute Deviation (MAD) to adaptively detect deviations, making it more resilient to outliers and distribution shifts. The LSTM-AE demonstrates superior performance in anomaly detection with respect to PCA and more recent model approaches, specifically for the reference dataset, obtained from a GlaxoSmithKline (GSK) production plant. Additionally, an LSTM regression model is employed to forecast future data points, which are then fed into the LSTM-AE to enable a predictive approach. This framework leverages the temporal dependencies captured by LSTM layers and reconstruction efficiency of the AE, facilitating a predictive anomaly detection in real-world applications.
Download

Paper Nr:	22
Title:	A Novel Automatic Monitoring and Control System For Induced Jet Breakup Fabrication of Ceramic Pebbles
Authors:	Miao Zhang, Oliver Leys, Markus Vogelbacher, Regina Knitter and Jörg Matthes
Abstract:	As the production of lithium-rich ceramic pebbles play a key role in the tritium-breeding blankets, it is vital for future fusion reactors. To ensure high-quality pebbles, the Karlsruhe Institute of Technology (KIT) has developed a melt-based fabrication process called KALOS (KArlsruhe Lithium OrthoSilicate). This process involves the break-up of a molten laminar jet to produce pebbles with precise diameters of hundreds of micrometers, which are highly dependent on process parameters. Therefore, a real-time monitoring and regulation system is essential for the fabrication process. This paper discusses a high-speed camera-based measurement system designed to automatically monitor and control the production process. Experimental evidence shows that this system can accurately provide real-time data on the sizes, locations, and distance distribution of the molten ceramic droplets utilizing image processing approaches. Additionally, the system is capable of controlling the production of pebbles by adjusting the driving frequency in real-time based on real-time measurements of the computer vision.
Download

Paper Nr:	40
Title:	Health-Aware Charging of Li-Ion Batteries Using MPC and Bayesian Degradation Models
Authors:	Taranjitsingh Singh, Jeroen Willems, Bruno Depraetere and Erik Hostens
Abstract:	We propose a Model Predictive Control (MPC) approach for health-aware optimal charging of Lithium-ion Nickel Manganese Cobalt (Li-NMC) batteries. Our method integrates electrical, thermal, and degradation models using Bayesian Networks (BNs) to estimate the battery’s State of Health (SOH). These models are embedded into an MPC framework to generate charging profiles that reduce long-term degradation while ensuring fast charging performance. Validation is performed through high-fidelity simulations using the PyBaMM battery modeling environment. Results show improved SOH retention compared to conventional Constant Current-Constant Voltage (CC-CV) strategy.
Download

Paper Nr:	47
Title:	Manipulation of Deformable Linear Objects Using Model Predictive Path Integral Control with Bidirectional Long Short-Term Memory Learning
Authors:	Lukas Zeh, Johannes Meiwaldt, Zexu Zhou, Armin Lechler and Alexander Verl
Abstract:	The manipulation of Deformable Linear Objects (DLOs) such as cables poses a significant challenge for automation due to their infinite degrees of freedom and non-linear dynamics. In this paper we present a machine learning based optimal control approach for the manipulation of DLOs. This approach is divided into two main components: modeling and control. For modeling the dynamics of the DLO, we propose a learning based approach using a bidirectional Long Short-Term Memory (biLSTM) network. The biLSTM network is trained on synthetic data generated by the MuJoCo physics engine. For manipulating the DLO, a model predictive control strategy that employs Model Predictive Path Integral (MPPI) control is selected. The proposed approach is evaluated through simulation and experiments. The results demonstrate the effectiveness of the proposed method in achieving accurate and efficient manipulation of DLOs.
Download

Paper Nr:	54
Title:	Reinforcement Learning for Model-Free Control of a Cooling Network with Uncertain Future Demands
Authors:	Jeroen Willems, Denis Steckelmacher, Wouter Scholte, Bruno Depraetere, Edward Kikken, Abdellatif Bey-Temsamani and Ann Nowé
Abstract:	Optimal control of complex systems often requires access to a high-fidelity model, and information about the (future) external stimuli applied to the system (load, demand, ...). An example of such a system is a cooling network, in which one or more chillers provide cooled liquid to a set of users with a variable demand. In this paper, we propose a Reinforcement Learning (RL) method for such a system with 3 chillers. It does not assume any model, and does not observe the future cooling demand, nor approximations of it. Still, we show that, after a training phase in a simulator, the learned controller achieves a performance better than classical rule-based controllers, and similar to a model predictive controller that does rely on a model and demand predictions. We show that the RL algorithm has learned implicitly how to anticipate, without requiring explicit predictions. This demonstrates that RL can allow to produce high-quality controllers in challenging industrial contexts.
Download

Paper Nr:	66
Title:	Approximating MPC Solutions Using Deep Neural Networks: Towards Application in Mechatronic Systems
Authors:	Edward Kikken, Jeroen Willems, Branimir Mrak and Bruno Depraetere
Abstract:	Model Predictive Control is an advanced control technique that can yield high performance, but it is often challenging to implement. Especially for systems with dynamics that are complex to model, have strong nonlinearities, and/or have small time constants, it is often not possible to complete the needed online optimizations fast and reliable enough. In this work we look at approximating the MPC solutions using black-box models i.e. deep neural networks, so that the computational load at runtime is strongly reduced. We use a supervised learning approach to train these models to yield outputs similar to those of an example dataset of offline pre-computed MPC solutions. We illustrate this approach on three realistic (active-suspension system, parallel robot, and a truck-trailer), illustrating the typical workflow and how the approach has to be set up to address the varying challenges. We show that the approximate MPC solutions yield a high level of performance, reaching nearly the level of the original MPC, yet at a strongly reduced computational load.
Download

Paper Nr:	73
Title:	Optimal Camera Placement for 6D Head Pose Estimation
Authors:	Harshita Soni, Nikhil Tirumala and Aratrik Chattopadhyay
Abstract:	Multi-view systems for 6D head pose estimation have applications in human-computer interaction (HCI), virtual reality, 3D reconstruction etc. In a multi-view system, visibility of facial landmarks is essential for accurately regressing 2D landmarks, which are then triangulated to get 3D fiducials. From these 3D fiducials, the 6D head pose is mathematically derived. Optimal camera placement (OCP) is vital for achieving precise pose estimation. OCP can be formulated as a constrained optimization problem that can be solved using Binary Integer Programming. We redefine two key aspects: the visibility criteria and the camera search space. Our visibility algorithm employs a parametric head model to track fiducials, achieving more precise results than ground truth of CMU(Carnegie Mellon University) Panoptic dataset. Additionally, we geometrically optimize the camera search space, deviating from the baseline of uniformly arranged cameras. Through rigorous experimentation, we prove that not only does this refined search space reduce execution time, but also improves the optimality of the solution, giving 99.9% visibility coverage. We also introduce a heuristic method that reduces the constraint-building time from 27 seconds to just 0.07 seconds per control point, while maintaining concise solutions with minimal effects on visibility metrics.
Download

Paper Nr:	80
Title:	Dual-Arm Manipulation of a T-Shirt from a Hanger for Feeding a Hem Sewing Machine
Authors:	Filipe Almeida, Gonçalo Leão, Carlos M. Costa, Cláudia D. Rocha, Armando Sousa, Lara Gomes da Silva, Luís F. Rocha and Germano Veiga
Abstract:	The textile industry is experiencing rapid advancement, reflected in the adoption of innovative and efficient manufacturing techniques. The automation of clothing sewing systems has the potential to reduce the allocation of repetitive tasks to operators, freeing them for more value-added operations. There are several machines on the market that automatically sew the bottom hem of T-shirts, a key component of the garment that fulfills both functional and aesthetic purposes. However, most of them require the fabric to be positioned manually by an operator. To address this issue, this work presents a solution to automate the process of feeding a T-shirt into a SiRUBA sewing machine using a YuMi dual-arm robot. In this scenario, the T-shirt arrives at the workstation with the main front and back pieces of cloth sewn together, seams facing out, and with no sleeves yet. This setup starts by turning the garment inside out with the aid of an automated hanger, ensuring that the seams are facing inward (as the machine requires), and then using the dual-arm robot to feed the garment into the sewing machine. With our approach, the feeding and hemming process took less than 35 seconds, with a feeding success rate of 98%. Therefore, this work can serve as a steppingstone towards more efficient automated sewing systems within the garment production industry.
Download

Paper Nr:	90
Title:	Solving the Three-Dimensional Beacon Placement Problem Using Constraint-Based Methods, Large Neighborhood Search, and Evolutionary Algorithms
Authors:	Sven Löffler, Viktoria Abbenhaus, George Assaf and Petra Hofstedt
Abstract:	With the increasing prevalence of large building complexes, indoor localization is becoming an area of growing significance. In critical situations, such as emergencies in factories or care facilities, the ability to locate a person quickly can be a matter of life and death. One possibility for localization are Bluetooth beacons, which are either attached to the person or in rooms. We pursue the latter approach, whereby the beacon signals are used to determine the position of a receiving device, e.g. a mobile phone. At this, the use of a sufficient number of beacons in the building must be ensured in order to guarantee adequate coverage. However, to minimize costs, it is equally important to avoid placing unnecessary beacons. This creates a challenging optimization problem that this paper addresses through three distinct approaches: constraint programming, large neighborhood search, and evolutionary algorithms. Using simulated three-dimensional buildings, we test and evaluate these methods, ultimately providing a practical and efficient approach applicable to real-world building environments.
Download

Paper Nr:	103
Title:	Place Recognition with Omnidirectional Imaging and Confidence-Based Late Fusion
Authors:	Marcos Alfaro, Juan José Cabrera, Enrique Heredia, Oscar Reinoso, Arturo Gil and Luis Paya
Abstract:	Place recognition is crucial for the safe navigation of mobile robots. Vision sensors are an effective solution to address this task due to their versatility and low cost, but the images are sensitive to changes in environmental conditions. Multi-modal approaches can overcome this limitation, but the integration of different sensors often leads to larger computing and hardware costs. Consequently, this paper proposes enhancing omnidirectional views with additional features derived from them. First, feature maps are extracted from the original omnidi-rectional images. Second, each feature map is processed by an independent deep network and embedded into a descriptor. Finally, embeddings are merged by means of a late approach that weights each feature according to the confidence in the prediction of the networks. The experiments conducted in indoor and outdoor scenarios revealed that the proposed method consistently improves the performance across different environments and lighting conditions, presenting itself as a precise, cost-effective solution for place recognition. The code is available at the project website: https://github.com/MarcosAlfaro/VPR LF VisualFeatures.
Download

Paper Nr:	114
Title:	Continual Multi-Robot Learning from Black-Box Visual Place Recognition Models
Authors:	Kenta Tsukahara, Kanji Tanaka, Daiki Iwata, Jonathan Tay Yu Liang and Wuhao Xie
Abstract:	In the context of visual place recognition (VPR), continual learning (CL) techniques offer significant potential for avoiding catastrophic forgetting when learning new places. However, existing CL methods often focus on knowledge transfer from a known model to a new one, overlooking the existence of unknown black-box models. This study explores a novel multi-robot CL approach that enables knowledge transfer from black-box VPR models (teachers), such as those of local robots encountered by traveler robots (students) in unknown environments. Specifically, we introduce Membership Inference Attack (MIA), a privacy attack applicable to black-box models, and leverage it to reconstruct pseudo training sets, which serve as the transferable knowledge between robots. Furthermore, we address the low sampling efficiency of MIA by leveraging prior insights from the literature on place class prediction distributions and unseen-class detection. Finally, we analyze both the individual and combined effects of these techniques.
Download

Paper Nr:	146
Title:	Enhancing Resilience of Strong Structural Controllability in Leader-Follower Networks
Authors:	Vincent Schmidtke and Olaf Stursberg
Abstract:	This paper explores measures of edge augmentation to enhance resilience of strong structural controllability for control systems modeled as leader-follower networks. Unlike existing methods which typically increase the number of leaders, the proposed approach achieves resilience by strategically adding edges, thus maintaining leader sets with a small cardinality. Using the zero forcing method, conditions are derived to enhance resilience either for specific agents or for the entire network. Numeric simulations validate the approach and show its effectiveness in large and complex networks.
Download

Paper Nr:	164
Title:	Predictive Quality of In-Fabrication Products in Smart Manufacturing Using Graph-Based Deep Learning
Authors:	Peter Davison, Muhammad Fahim, Roger Woods, Scott Fischaber, Marcus Heron and Cormac McAteer
Abstract:	Graph neural networks are a very powerful way to learn about relationships between entities in graphs. With the rise of IoT devices in manufacturing, more data is being collected to minimise the waste of both valuable resources and time for fabrication. In this paper, we introduce a methodology for predictive quality of in-fabrication products using graph neural networks. Data is collected from a live-working semiconductor wafer fabrication facility and used to produce heterogeneous graphs that represent the fabrication timeline of a wafer. The model uses the graph attention network architecture to classify whether a timeline is scrap or non-scrap. It uses historical graph-level labelled data and achieves an F1-score of 0.928, compared to baselines models of a LSTM and a Homogeneous Graph Attention Network with scores of 0.424 and 0.786 respectively. It gives a foundational framework for future anomaly detection with semiconductor fabrication, allowing real-world data to be analysed with graph-based deep learning tools to provide interpretation and accessible graph-based results.
Download

Paper Nr:	167
Title:	Application of MPPT Techniques Using Intelligent and Conventional Control Strategies
Authors:	João T. Sousa and Ramiro S. Barbosa
Abstract:	This paper presents a comparative study of five MPPT (Maximum Power Point Tracking) algorithms applied to photovoltaic (PV) systems under both uniform and dynamic environmental conditions. The analyzed algorithms include two conventional methods, Perturb \& Observe (P\&O) and Incremental Conductance (InC), as well as a fuzzy logic controller (FLC) and two hybrid strategies enhanced by genetic algorithms (P\&O+GA and InC+GA). A unified simulation framework in MATLAB/Simulink was used to ensure fair benchmarking, employing identical panel configurations, irradiance/temperature profiles, and converter parameters. Each algorithm was tested using predefined parameters such as step size, initial duty cycle, and operating bounds. Additionally, an EMA (Exponential Moving Average) filter was applied to the hybrid algorithms to reduce high-frequency measurement noise. Evaluation metrics include Mean Absolute Error (MAE), Integral Absolute Error (IAE), Mean Squared Error (MSE), Integral Squared Error (ISE), convergence time, and energy conversion efficiency. Results demonstrate that hybrid methods deliver superior performance in noisy and fast-changing conditions, while FLC maintains stable performance with reduced oscillations. This work aims to support the selection of suitable MPPT techniques for real-world PV systems, balancing computational complexity and control effectiveness.
Download

Short Papers

Paper Nr:	30
Title:	Leader-Follower Coordination in UAV Swarms for Autonomous 3D Exploration via Reinforcement Learning
Authors:	Robert Kathrein, Julian Bialas, Mohammad Reza Mohebbi, Simone Walch, Mario Döller and Kenneth Hakr
Abstract:	Autonomous volumetric scanning in three-dimensional environments is critical for environmental monitoring, infrastructure inspection, and search and rescue applications. Efficient coordination of multiple Unmanned Aerial Vehicles (UAVs) is essential to achieving complete and energy-aware coverage of complex spaces. In this work, a Reinforcement Learning (RL)-based framework is proposed for the coordination of a leader-follower UAV system performing volumetric scanning. The system consists of two heterogeneous UAVs with directional sensors and constant mutual orientation during the mission. A centralized control policy is learned based on Proximal Policy Optimization (PPO) to control the leader UAV, which produces trajectory commands for the follower to achieve synchronized movement and effective space coverage. The observation space includes a local 3D occupancy map of the leader and both UAVs’ battery levels, enabling energy-aware decision-making. The reward function is carefully designed to favor exploration and visiting new regions without penalizing collision and boundary crossing. The proposed method is verified using both simulation experiments and real-world experiments on the ArduPilot platform, showing the applicability of RL to scalable autonomous multi-UAV scanning operations.
Download

Paper Nr:	34
Title:	Mobile Application with Convolutional Neural Networks for the Early Detection of Diseases in Blueberry Plants in Chepén: Trujillo
Authors:	Santiago Sebastian Heredia Orejuela and Aaron Moises Cosquillo Garay
Abstract:	Early detection of foliar diseases in blueberry crops is essential to protect yield and fruit quality, especially in Chepén–Trujillo, a key agricultural region in Peru. This paper presents a mobile application developed with Flutter and powered by a lightweight convolutional neural network (CNN), capable of analyzing leaf images and delivering disease diagnoses in under three seconds. The system supports offline functionality, ensuring usability in rural areas with limited connectivity. In a test set of 350 images, the model achieved 93% accuracy, 88% recall, and an F1 score of 0.90. Field validation with local farmers showed 90% agreement with expert diagnoses. Beyond its technical performance, this solution has the potential to reduce economic losses, improve crop quality, and empower smallholder farmers through accessible, real-time diagnostics. The platform is scalable to other crops and regions, contributing to more sustainable and resilient agricultural practices in Peru.
Download

Paper Nr:	46
Title:	Observation-Based Inverse Kinematics for Visual Servo Control
Authors:	Daniel Nikovski
Abstract:	We propose a method for estimating the joint configuration of articulated mechanisms without joint encoders and with unknown forward kinematics, based solely on RGB-D images of the mechanism captured by a stationary camera. The method collects a sequence of such images under a suitable excitation control policy, extracts the 3D locations of keypoints in these images, and determines which of these points must belong to the same link of the mechanism by means of testing their pairwise distances and clustering them using agglomerative clustering. By computing the rigid-body transforms of all bodies with respect to the keypoints’ positions in a reference image and analyzing each body’s transform expressed relative to all other bodies’ coordinate reference frames, the algorithm discovers which pairs of bodies must be connected by a single-degree-of-freedom joint and based on this, discovers the ordering of the bodies in the kinematic chain of the mechanism. The method can be used for pose-based visual servocontrol and other robotics tasks where inverse kinematics is needed, without providing forward kinematics or measurements of the end tool of the robot.
Download

Paper Nr:	49
Title:	Evaluation Approaches for an Aggregated Meteorological Model for Artillery Operations
Authors:	Jan Ivan, Viktor Vitoul, Ladislav Potužák and Jan Drábek
Abstract:	This article presents a research initiative focused on developing innovative methods of meteorological preparation for artillery units. The ongoing conflict in Ukraine has underscored the pivotal role of artillery for both sides, with its effectiveness hinging on fire accuracy-requiring compensation for multiple variables affecting projectile trajectory. Among these variables, meteorological conditions are paramount and have traditionally been assessed via upper-atmosphere sounding. However, current methods are susceptible to enemy interference, necessitating the autonomous acquisition of meteorological data by artillery units, even under degraded operational conditions. This research project proposes the development of an integrated predictive model that leverages historical meteorological data. Using this model, artillery units would be able to independently generate meteorological insights, eliminating the need for complex atmospheric sounding systems or reliance on external data sources. The article also outlines a proposed method for evaluating the model’s effectiveness, based on the General Preparation procedure used in artillery fire control.
Download

Paper Nr:	61
Title:	Vision-Based Autonomous Landing for the MPC Controlled Fixed Wing UAV
Authors:	Sevinç Günsel, Şeref Naci Engin and Mustafa Doğan
Abstract:	This work introduces a novel vision-based autonomous landing system for fixed-wing UAVs optimized for GPS-denied environments. We combine vSLAM with the linear MPC strategy. A key innovation is to use an SVD-based Kalman filter in vSLAM, which significantly improves map point update accuracy and efficiency by reducing noise. The system precisely defines the landing area using image segmentation and Watershed Transform for real-time vSLAM data, then draws a rotated bounding box. This visual data feeds the linearized MPC, which computes the optimal control inputs which are longitudinal acceleration, yaw rate, vertical velocity to guide the UAV along the landing trajectory. Simulation results confirm the robust and effective performance of our integrated vSLAM-MPC architecture in precisely guiding the UAV to the landing zone.
Download

Paper Nr:	64
Title:	Enhancing PI Tuning for Plant Commissioning Using Transfer Learning and Bayesian Optimization
Authors:	Boulaid Boulkroune, Joachim Verhelst, Branimir Mrak, Bruno Depraetere, Joram Meskens and Pieter Bovijn
Abstract:	A novel approach for accelerating the auto-tuning of PI controllers during the commissioning phase is proposed in this study. This approach combines transfer learning and Bayesian optimization (BO) to minimize the number of iterations required to converge to the optimal solution. Transfer learning is employed to extract valuable information from available historical data derived from expert tuning of other equivalent process variants. In the absence of historical data, a simulation model can also be utilized to generate data from different model variants (e.g., changing the value of unknown parameters). In this study, a simulation model is used for generating historical data. The approach’s efficiency is demonstrated through its application to a thermal plant, achieving a significant reduction in the number of iterations required to reach the optimizer’s optimal solution.
Download

Paper Nr:	68
Title:	Towards Scalable and Fast UAV Deployment
Authors:	Tim Felix Lakemann and Martin Saska
Abstract:	This work presents a scalable and fast method for deploying Uncrewed Aerial Vehicle (UAV) swarms. Decentralized large-scale aerial swarms rely on onboard sensing to achieve reliable relative localization in real-world conditions. In heterogeneous research and industrial platforms, the adaptability of the individual UAVs enables rapid deployment in diverse mission scenarios. However, frequent platform reconfiguration often requires time-consuming sensor calibration and validation, which introduces significant delays and operational overhead. To overcome this, we propose a method that enables rapid deployment and calibration of vision-based UAV swarms in real-world environments.
Download

Paper Nr:	94
Title:	Quantitative Analysis of Ambient Temperature Effects on Steptime Variations in Industrial Pneumatic Actuators
Authors:	Jon Zubieta, Unai Izagirre and Luka Eciolaza
Abstract:	This paper presents a quantitative analysis of the influence of ambient temperature on the cycle time of pneumatic actuators in industrial production environments. Sub-cycle time periods, known as Steptimes, are used to characterize the duration of individual machine stages without requiring additional sensors. Building on the concept of Mini-terms and following the IEC 60848 GRAFCET standard, Steptimes are defined as the elapsed time between the activation and deactivation of PLC-controlled steps. Although the potential impact of ambient temperature on actuator performance is often acknowledged qualitatively, few studies have addressed this effect through precise, quantitatively measured data. In this work, a detailed experimental study is conducted using a PLC-controlled system composed of four automated modules. Steptimes and ambient temperature have been continuously monitored and their effects modeled statistically. The results show a consistent inverse correlation between temperature and Steptimes, as expected. The contribution of this research work is twofold: first, the feasibility and potential of using Steptime measurements to detect subtle environmental effects in industrial assembly lines is demonstrated. Second, the impact of ambient temperature in highly automated industrial assembly lines is quantitatively measured. By modeling subtle environmental effects, deviations in Steptime can be more accurately interpreted, reducing the risk of false alarms and improving system reliability.
Download

Paper Nr:	100
Title:	Data-Driven Control of a PEM Electrolyzer
Authors:	Yeyson A. Becerra-Mora, Juan Manuel Escaño and José Ángel Acosta
Abstract:	Green hydrogen production has gained significant relevance in recent years to substitute fossil fuels in the coming years. One of the most promising technologies for attaining such a milestone is the PEM electrolyzer; nevertheless, some considerations related to controlling its temperature must be addressed, such as avoiding high temperatures to extend its useful life and improve its efficiency. Therefore, this study proposes a data-driven control strategy based on Gaussian Process Regression (GPR) and Nonlinear Model Predictive Control (NMPC). GPR is used to identify the system, while NMPC is used to regulate the output temperature of the PEM electrolyzer with the identified model. Simulations show a clear resemblance between the Gaussian Process model and the phenomenological model, as well as the effectiveness of the controller. Furthermore, error metrics and computational time are presented.
Download

Paper Nr:	101
Title:	An Educational Platform for Real-Time Control and Reinforcement Learning Experiments Using Rotary Inverted Pendulum and LW-RCP
Authors:	Doyoon Ju, Jongbeom Lee and Young Sam Lee
Abstract:	This paper presents an integrated experimental platform for hands-on education in control engineering, built around a compact rotary inverted pendulum system based on a stepper motor and a Simulink-based Light Weight Rapid Control Prototyping (LW-RCP) environment. The proposed platform supports real-time implementation of a wide range of control experiments, including current-based vector control, nonlinear swing-up control, linear stabilization, and reinforcement learning-based control. The hardware consists of a rotary inverted pendulum made with 3D-printed components and a hollow-shaft stepper motor, incorporating a compact inverter realized through an L6234 motor driver. Its compact and lightweight design allows for tabletop experimentation, enabling one-device-per-student operation and enhancing scalability in education. On the software side, users can design controllers and collect real-time data through Simulink’s block-based modeling interface without coding. Moreover, Python integration enables sim-to-real experiments with reinforcement learning controllers. This platform complements traditional theory-centric control engineering education by offering rich hands-on experiences, thereby increasing student motivation and fostering a deeper conceptual understanding through the full process of controller design and system response analysis.
Download

Paper Nr:	105
Title:	Hierarchical Coordination of UAVs for Dynamic Task Assignment in Large-Scale Traffic Surveillance Missions
Authors:	Teewende Boris kiema, Hélène Piet-Lahanier, Najett Neji and Samia Bouchafa
Abstract:	This paper presents an hierarchical coordination architecture for a fleet of UAVs dedicated to road traffic surveillance over large urban areas. The system is built around a central drone, acting as a coordinator, which is responsible for monitoring the status of the fleet and dynamically assigning surveillance tasks in response to reported traffic events. To ensure scalability and responsiveness, our architecture combines a spatial clustering mechanism to partition mission area and distribute drones accordingly, with a receding horizon task assignment (RHTA) strategy within each sub-region. The fleet coordination requires designing specific trajectories for the central drone to ensure communication within the fleet and periodic updates of the surveillance information. This hybrid approach enables adaptive, region-based task allocation while preserving a global overview through the coordinator. Simulation results highlight the relevance and flexibility of the proposed coordination scheme when addressing dynamic and large-scale surveillance scenarios.
Download

Paper Nr:	113
Title:	A Robust Comparative Study of Adaptative Reprojection Fusion Methods for Deep Learning Based Detection Tasks with RGB-Thermal Images
Authors:	Enrique Heredia-Aguado, Marcos Alfaro-Pérez, María Flores, Luis Paya, David Valiente and Arturo Gil
Abstract:	Fusing visible and thermal imagery is a promising approach for robust object detection in challenging environments, taking advantage of the strengths from different spectral information. Building on previous work in static early fusion, we present a comparative study of adaptative reprojection fusion methods that exploit advanced projections and frequency-domain transforms to combine RGB and thermal data. We evaluate Principal Component Analysis, Factor Analysis, Wavelet and Curvelet-based fusion, all integrated into a YOLOv8 detection pipeline. Experiments are conducted on the LLVIP dataset, with a focus on methodological rigour and reproducibility. This research show promising results based on these methods comparing to previous early fusion methods. We discuss the implications for future research and the value of robust experimental design for advancing the state of the art in multispectral fusion.
Download

Paper Nr:	120
Title:	Balancing Speed and Accuracy: A Comparative Analysis of Segment Anything-Based Models for Robotic Indoor Semantic Mapping
Authors:	Bruno Georgevich Ferreira, Armando Jorge Sousa and Luis Paulo Reis
Abstract:	Semantic segmentation is a relevant process for creating the rich semantic maps required for indoor navigation by autonomous robots. While foundation models like Segment Anything Model (SAM) have significantly advanced the field by enabling object segmentation without prior references, selecting an efficient variant for real-time robotics applications remains a challenge due to the trade-off between performance and accuracy. This paper evaluates three such variants - FastSAM, MobileSAM, and SAM 2 - comparing their speed and accuracy to determine their suitability for semantic mapping tasks. The models were assessed within the Robot@VirtualHome dataset across 30 distinct scenes, with performance quantified using Frames Per Second (FPS), Precision, Recall, and an Over-Segmentation metric, which quantifies the fragmentation of an object into multiple masks, preventing high quality semantic segmentation. The results reveal distinct performance profiles: FastSAM achieves the highest speed but exhibits poor precision and significant mask fragmentation. Conversely, SAM 2 provides the highest precision but is computationally intensive for real-time applications. MobileSAM emerges as the most balanced model, delivering high recall, good precision, and viable processing speed, with minimal over-segmentation. We conclude that MobileSAM offers the most effective trade-off between segmentation quality and efficiency, making it a good candidate for indoor semantic mapping in robotics.
Download

Paper Nr:	128
Title:	Recursive Gaussian Process Regression with Integrated Monotonicity Assumptions for Control Applications
Authors:	Ricus Husmann, Sven Weishaupt and Harald Aschemann
Abstract:	In this paper, we present an extension to the recursive Gaussian Process (RGP) regression that enables the satisfaction of inequality constraints and is well suited for a real-time execution in control applications. The soft inequality constraints are integrated by introducing an additional extended Kalman Filter (EKF) update step using pseudo-measurements. The sequential formulation of the algorithm and several developed heuristics ensure both the performance and a low computational effort of the algorithm. A special focus lies on an efficient consideration of monotonicity assumptions for GPs in the form of inequality constraints. The algorithm is statistically validated in simulations, where the possible advantages in comparison with the standard RGP algorithm become obvious. The paper is concluded with a successful experimental validation of the developed algorithm for the monotonicity-preserving learning of heat transfer values for the control of a vapor compression cycle evaporator, leveraging a previously published partial input output linearization (IOL).
Download

Paper Nr:	129
Title:	Real-Time Fault Detection and Diagnosis for Oil Well Drilling Using a Multitask Neural Network
Authors:	Marios Gkionis, Ole Morten Aamo and Ulf Jakob Flø Aarsnes
Abstract:	Drilling operations can be unexpectedly laden with mechanical faults, mud loss, and insufficient cuttings transport that incur substantial costs. This can be avoided via accurate and early fault detection and diagnosis. We present a novel Drilling Fault Detection and Diagnosis (FDD) system that leverages Multitask Neural Networks (MTL-NNs). It accounts for the practical limitation that down-hole measurements are normally not available in real-time and can perform FDD relying only on flow and pressure measurements at the drilling rig. Data for training and testing are produced by a simulator based on the distributed flow and pressure dynamics in the entire well governed by four coupled hyperbolic partial differential equations. Faults are incorporated into the simulations so that the data contain information about how diagnostics of faults affect the dynamics. Our numerical experiments, admittedly under quite ideal conditions, show that the proposed method exhibits high generalization performance on diagnosis for fixed well depths, while incorporating varying well depths into a single network requires increased size in both network and training data to maintain performance.
Download

Paper Nr:	153
Title:	Adaptive Trajectory Prediction in Roundabouts Using Moving Horizon Estimation
Authors:	Selsabil Bougherara, Hasni Arezki, Chouki Sentouh, Jérôme Floris and Jean-Christophe Popieul
Abstract:	This paper addresses the challenge of trajectory prediction for automated vehicles navigating within roundabouts, where interactions, non-linear motion, and rapid decision-making complicate traditional approaches. We propose a novel prediction framework based on Moving Horizon Estimation (MHE) combined with a nonlinear kinematic bicycle model. Unlike conventional methods such as the Extended Kalman Filter (EKF), the proposed MHE-based framework leverages past observations over a sliding time window, enhancing robustness against model uncertainties and noise. The method is validated through simulations using the SHERPA driving simulator in both static and dynamic maneuvering. The results demonstrate that MHE significantly outperforms the EKF in terms of prediction accuracy, particularly during complex vehicle behaviors. This work constitutes a foundational step toward enhancing safety and robustness of decision-making in roundabouts.
Download

Paper Nr:	17
Title:	Time-Optimal Scheduling of Tasks with Shared and Dynamically Constrained Energy Systems
Authors:	Eero Immonen
Abstract:	This article addresses the minimum-time scheduling of sequential tasks requiring energy (or a similar resource) from shared, dynamically constrained systems. Practical applications of this problem include human operations with fatigue and rest cycles, among others. The goal is to jointly optimize task execution order and power allocation to the tasks, balancing execution speed with necessary recovery periods and task transition times. We present a generic Mixed-Integer Nonlinear Programming (MINLP) formulation of the problem, propose a heuristic solution method based on a Genetic Algorithm (GA), and demonstrate its use in a numerical example on efficient execution of a two-exercise workout. The numerical example shows that the proposed heuristic method rapidly produces a solution within 0.9% of the one obtained via the MINLP solver SCIP.
Download

Paper Nr:	31
Title:	Translating NWP Outputs into UAV-Specific Predictions Using Machine Learning
Authors:	David Sládek
Abstract:	Unmanned Aerial Vehicles (UAVs) are increasingly deployed in safety-critical, weather-sensitive operations. However, the direct use of Numerical Weather Prediction (NWP) model outputs often fails to address the specific operational thresholds and spatial–temporal needs of UAV missions. This study introduces a machine learning (ML) framework that translates standard NWP forecasts into UAV-specific feasibility assessments. We integrate both global (GFS) and local high-resolution (ARPEGE, AROME) models to generate real-time, interpretable indices or GO/NO-GO indicators tailored to UAV performance limits. Our case study over Nantes (France) for the 2017–2023 period demonstrates the added value of ML-enhanced predictions in terms of spatial precision, temporal consistency, and decision-support utility. The proposed approach also offers an effective method to fill gaps in local model availability by learning from global models, ensuring continuity and operational resilience. By combining observation statistics, NWP forecasts, and ML interpretation, this methodology supports scalable, automated pre-flight planning under varying weather scenarios.
Download

Paper Nr:	50
Title:	Categorical Model Estimation with Feature Selection Using an Ant Colony Optimization
Authors:	Tetiana Reznychenko, Evženie Uglickich and Ivan Nagy
Abstract:	This paper deals with the analysis of high-dimensional discrete data values from questionnaires, with the aim of identifying explanatory variables that influence a target variable. We propose a hybrid algorithm that combines categorical model estimation with an ant colony optimization scheme for feature selection. The main contributions are: (i) the efficient selection of the most significant explanatory variables, and (ii) the estimation of a categorical model with reduced dimensionality. Experimental results and comparisons with well-known algorithms (e.g., random forest, categorical boosting, k-nearest neighbors) and feature selection techniques are presented.
Download

Paper Nr:	71
Title:	Real-Time Weld Quality Prediction in Automated Stud Welding: A Data-Driven Approach
Authors:	Beatriz Coutinho, Bruno Santos, Rita Gomes Mendes, Gil Gonçalves and Vítor H. Pinto
Abstract:	Drawn arc stud welding is extensively used in automotive assembly lines for attaching components to vehicle bodies. In these automated processes, low-quality welds can compromise structural integrity and cause production delays due to rework and maintenance. This paper describes the initial development stage of an artificial intelligence (AI)-based system for real-time weld quality prediction in automated stud welding. The focus of this first phase is on implementing sensorisation, developing a data acquisition system, and constructing a dataset that captures the most relevant process variables characterizing the welding process. A Flask-based application was developed to facilitate data collection, incorporating an automatic character recognition algorithm to extract parameters directly from the control unit display. Initial welding experiments produced a dataset of approximately 200 samples, with preliminary data analysis validating expected parameter trends. The results confirm the system’s capability to effectively capture relevant data, forming the basis for future development of a predictive model aimed at enhancing weld quality monitoring and minimizing assembly line interruptions.
Download

Paper Nr:	104
Title:	Automated Process Control for the Beam Gas Curtain Vacuum System at CERN
Authors:	L. Cantu, R. Ferreira, J. Francisco Rebelo, A. Rocha, C. Vazquez Pelaez and L. Zygaropoulos
Abstract:	The Beam Gas Curtain (BGC) system is a key diagnostic instrument for non-invasive proton beam profiling in the Large Hadron Collider (LHC), relying on precise and safe gas injection into the beam pipe. Initially operated via manual procedures through a supervisory control and data acquisition (SCADA) interface, BGC injections required expert users, were time consuming and vulnerable to human error. This paper presents the design and implementation of an automated gas injection control system, fully integrated within the LHC Vacuum Control System SCADA and using Vacuum Framework. The solution includes a finite state machine (FSM) deployed on a programmable logic controller (PLC), a new state-aware SCADA interface, and a comprehensive interlock strategy combining device-level and process-level safety. The system was extensively tested using simulations and staged commissioning, culminating in a successful deployment during the LHC Year-End Technical Stop (YETS) 2024/25. Automation has drastically simplified operations, increased reliability, and enhanced machine safety, requiring only two user actions to initiate an injection.
Download

Paper Nr:	108
Title:	Intelligent Process Automation Model for Credit Campaign Management Optimization in a Financial Institution
Authors:	Fernando Schilder Hervias and Neil Trujillo Nerya
Abstract:	In Peru, financial institutions face significant challenges in credit campaign management due to inefficient operational processes dependent on traditional procedures. This generates delays, limits their response capacity, and reduces their competitiveness. To address these challenges, the present study has proposed an Intelligent Process Automation (IPA) model as an innovative solution to optimize credit campaign management. The model integrates Robotic Process Automation (RPA) and Artificial Intelligence (AI) technologies, along with an administration console for control and monitoring. We validated the proposal through a case study in a banking entity located in Lima, Peru, where these technologies had not been previously implemented. The results show significant operational improvements: execution time reduction of up to 85%, error rates below 1.5%, and a potential economic impact of S/ 363,000-484,000 (peruvian soles) monthly, validating the effectiveness of the proposed model for optimizing credit management. This study contributes to the emerging field of intelligent automation in financial services and provides a model for future implementations in the sector.

Paper Nr:	111
Title:	Smart Water Management: Integrating PLC and SCADA Technologies for Sustainable Urban Infrastructure
Authors:	Nirmal Kumar Balaraman, Krunal Patel and Nagender Reddy
Abstract:	This paper explores the integration of automation technologies-Programmable Logic Controllers (PLCs) and Supervisory Control and Data Acquisition (SCADA) systems-in advancing sustainable water and wastewater management within smart buildings and urban infrastructure. Two case studies are analysed. The first one focuses on a PLC-based automatic water distribution system for smart residential buildings, aligned with the United Nations Sustainable Development Goals (SDGs), highlighting its ability to reduce water waste, improve billing accuracy, and promote responsible consumption. The second one examines a SCADA-monitored water storage and distribution network in Yozgat province, Turkey, demonstrating significant reductions in water loss and improvements in water quality monitoring through centralised, real-time control. The discussion addresses the implications of integrating IoT, PLC, and SCADA technologies in water automation, emphasising enhanced efficiency, system intelligence, and the potential for predictive maintenance. Scalability across urban and rural contexts and broader industry applications is also considered. Despite the promise of automation, challenges such as high initial costs, data integration complexities, and skill gaps are identified, with recommendations for future implementation strategies. The study concludes that smart water management solutions are essential for sustainable urban development, offering actionable insights for policymakers, engineers, and city planners aiming to create resilient, data-driven water infrastructures.
Download

Paper Nr:	124
Title:	Satellite Navigation Constellation Optimisation Problem Definition for the Application of Genetic Algorithms
Authors:	Paula Piñeiro Ramos, Sebastian Bernhardt, Helena Stegherr and Jörg Hähner
Abstract:	Global Navigation Satellite Systems (GNSS) are used on a daily basis, providing Positioning, Navigation and Timing (PNT) services for various applications ranging from smartphones over the financial sector up to areas such as aviation and space. Classical GNSS constellations positioned in Medium Earth Orbit (MEO) often experience reduced performance in areas of low visibility like forests and cities. To rectify this, augmentation constellations are deployed, improving the provided positioning accuracy. Recent proposals for augmentation systems have often been based in Low Earth Orbit (LEO), which, for global coverage, require a large number of satellites and are complex to design due to dependencies, coverage requirements and the large search space. This makes the constellation design problem well-suited for applying Genetic Algorithms (GA) to find an optimal solution. However, previous research has only addressed highly constrained versions of the problem. This paper presents an approach for applying GAs to constellation designs with a large search space. In particular, the focus is on the description of the multi-objective fitness function and the simulation necessary for its evaluation, options for the solution encoding, and a discussion of algorithmic features applicable in this scenario.
Download

Paper Nr:	135
Title:	Improving Assistive Technologies Using EEG Headsets
Authors:	David Ivaşcu and Isabela Drămnesc
Abstract:	Brain computer interfaces (BCI) have gained increasing attention in recent years due to the improved afford-ability and usability of electroencephalogram headsets (EEG). These headsets paired with the right software make computers usable without a physical input such as the traditional mouse and keyboard, creating new opportunities for users with motor impairments. In this paper, we present the design and development of an assistive application that employs an EEG headset (Unicorn Hybrid Black) as the main control interface for user interaction. The system integrates a launcher style interface that contains multiple accessible functions, allowing users to interact with software environments exclusively through EEG–based commands. This work aims to advance digital accessibility and promote independence for people who cannot rely on conventional input devices. By outlining a practical approach for integrating EEG headsets into everyday computer use, this paper contributes to the ongoing development of assistive technologies.
Download

Paper Nr:	145
Title:	Kalman Type Filtering in the Presence of Parametric Modeling Uncertainties: A Navigation System Application for Launch Vehicles
Authors:	Adrian-Mihail Stoica
Abstract:	The paper presents a Kalman type filtering problem for a class of discrete-time stochastic systems corrupted with state dependent noises. The filter gain is derived solving an H2 optimization problem for the estimation error and it is expressed in terms of the solution of a specific system of coupled Riccati and Lyapunov equations. The design procedure is illustrated for a navigation system of a launch vehicle which design model includes parametric uncertainties.

Paper Nr:	148
Title:	Pedestrian Positioning Technology Combining IMU and Wireless Signals Based on MC-CKF
Authors:	Seong Yun Cho and Jae Uk Kwon
Abstract:	In the paper, the pedestrian position is estimated by integrating the inertial measurement unit (IMU) and the wireless signal using the Cubature Kalman filter (CKF) based on the maximum correntropy criterion (MCC). Wireless signals may include short-range wireless communications such as ultra-wideband (UWB) signal and mobile communication signals such as LTE/5G. UWB can measure distances with an error of less than 30 cm in a line-of-sight (LoS) environment, but in an environment with LoS, it provides range measurements with a wide range of non-Gaussian uncertainty errors. In this case, ia an IMU/UWB system is configured with a conventional minimum mean square error (MMSE)-based filter, significant errors will occur. To address this issue, this paper designed an MCC-based CKF and applied it to pedestrian positioning technology. Simulation analysis results demonstrated that the proposed filter is robust to UWB uncertainty and enables reliable IMUUWB integration.
Download

Paper Nr:	155
Title:	Combining off-Policy and on-Policy Reinforcement Learning for Dynamic Control of Nonlinear Systems
Authors:	Ahmed A. Hani Hazza, Simon G. Fabri, Marvin K. Bugeja and Kenneth. Camilleri
Abstract:	This paper introduces QARSA, a novel reinforcement learning algorithm that combines the strengths of off-policy and on-policy methods, specifically Q-learning and SARSA, for the dynamic control of nonlinear systems. Designed to leverage the sample efficiency of off-policy learning while preserving the stability and lower variance of on-policy approaches, QARSA aims to offer a balanced and robust learning framework. The algorithm is evaluated on the CartPole-v1 simulation environment using the OpenAI Gym framework, with performance compared against standalone Q-learning and SARSA implementations. The comparison is based on three critical metrics: average reward, stability, and sample efficiency. Experimental results demonstrate that QARSA outperforms both Q-learning and SARSA, achieving higher average rewards, stability, sample efficiency, and improved consistency in learned policies. These results demonstrate QARSA’s effectiveness in environments were maximizing long-term performance while maintaining learning stability is crucial. The study provides valuable insights for the design of hybrid reinforcement learning algorithms for continuous control tasks.
Download

Area 3 - Robotics and Automation

Full Papers

Paper Nr:	23
Title:	Sensorless Admittance Control of a Manipulator Arm Using a Nonlinear Observer for Force and Velocity Estimation
Authors:	Brahim Brahmi
Abstract:	This paper presents a nonlinear observer-based approach for estimating force and velocity in a joint-space admittance-controlled exoskeleton, designed to support safe and compliant physical human–robot interaction. The observer estimates external interaction forces and joint velocities using only joint position measurements, eliminating the need for external force or velocity sensors. Integrated into the ETS-MARSE upper-limb rehabilitation exoskeleton, the system generates compliant motion trajectories in response to user-applied forces. An experiment involving a human subject was conducted to validate the observer’s accuracy. The estimated forces and velocities were compared with reference sensor measurements. Results demonstrate that the observer provides reliable state estimates, enabling accurate tracking of motion and interaction forces with low error and high responsiveness. The system maintains compliant behavior, supporting natural, user-driven movement without compromising stability. This work highlights the potential of sensorless estimation in robotic rehabilitation and interaction-intensive control applications.
Download

Paper Nr:	24
Title:	ALEX-GYM-1: A Novel Dataset and Hybrid 3D Pose Vision Model for Automated Exercise Evaluation
Authors:	Ahmed Hassan, Abdelaziz Serour, Ahmed Gamea and Walid Gomaa
Abstract:	Improper gym exercise execution often leads to injuries and suboptimal training outcomes, yet conventional assessment relies on subjective human observation. This paper introduces ALEX-GYM-1, a novel multi-camera view dataset with criterion-specific annotations for squats, lunges, and Romanian deadlifts, alongside a complementary multi-modal architecture for automated assessment. Our approach uniquely integrates: (1) a vision-based pathway using 3D CNN to capture spatio-temporal dynamics from video, and (2) a pose-based pathway that analyzes biomechanical relationships through engineered landmark features. Extensive experiments demonstrate the superiority of our Multi-Modal fusion architecture over both single-modality methods and competing approaches, achieving Hamming Loss reductions of 30.0% compared to Vision-based and 79.5% compared to Pose-based models. Feature-specific analysis reveals key complementary strengths, with Vision-based components excelling at contextual assessment (89% error reduction for back knee positioning) while Pose-based components demonstrate precision in specific joint relationships. The computational efficiency analysis enables practical deployment strategies for both real-time edge applications and high-accuracy cloud computing scenarios. This work addresses critical gaps in exercise assessment technology through a purpose-built dataset and architecture that establishes a new state-of-the-art for automated exercise evaluation in multi-camera view settings.
Download

Paper Nr:	28
Title:	Evaluation of YOLO Architectures for Automated Transmission Tower Inspection Under Edge Computing Constraints
Authors:	Gabriel Jose Scheid, Ronnier Frates Rohrich and André Schneider de Oliveira
Abstract:	This paper explores YOLO architectures for the automated inspection of transmission towers using drone imagery, focusing on edge computing constraints. The approach assesses various model variants on a specialized dataset, optimizing their deployment on embedded hardware through strategic core allocation and format conversion. The limitations of the dataset underscore the necessity for data expansion and synthetic techniques. In addition, practical guidelines address the trade-offs between computational resources and performance in energy monitoring. Our approach aims to ensure reliable obstacle classification in cameras designed for robotic vision by mimicking human perception. The sensor combines stereo depth and high-resolution color cameras with on-device Neural Network inferencing and Computer Vision capabilities, all integrated into a single portable sensor suitable for use in autonomous Unmanned Aerial Vehicles (UAVs).
Download

Paper Nr:	35
Title:	A Depth Image Processing Algorithm for Monitoring Product Flow on a Conical Feeder Unit of a Multihead Weigher
Authors:	Julia Isabel Hartmann and Christoph Ament
Abstract:	Cameras are widely used as sensors in both industrial and research settings for tasks such as quality inspection, measurement, process monitoring, and control. Depending on the application, customized image processing algorithms are required to extract quantitative measurement data from captured images. This paper presents a novel depth image-based data acquisition algorithm for tracking the motion of multiple products on the rotating conical feeder unit of a multihead weigher, an industrial weighing system composed of multiple load cells. The acquired depth image data can be used for the parameter identification and verification of a model, which simulates particle motion on the feeder surface. Future work aims at implementing the image processing algorithm on a programmable logic controller to enable real-time tracking and integration with the control system of the multihead weigher in an industrial environment.
Download

Paper Nr:	36
Title:	Towards Guaranteed Collision Avoidance for Multiple Autonomous Underactuated Unmanned Surface Vehicles in Restricted Waters
Authors:	Erick J. Rodríguez-Seda
Abstract:	As autonomous surface vessels increasingly operate in restricted and congested waters, the need for distributed, reactive collision avoidance algorithms becomes more crucial. Traditional avoidance control algorithms are typically conservative, opting for a worst-case scenario approach and restricting the total area where Unmanned Surface Vehicles (USVs) can navigate. This paper presents a distributed collision avoidance framework for USVs, based on the concepts of Artificial Potential Field (APF) and avoidance functions, that aims to reduce the minimum safe distance that vehicles need to keep from obstacles by explicitly considering their shape, relative position, and relative orientation. The proposed control framework is theoretically demonstrated and validated through simulations to ensure collision avoidance at all times and to facilitate the travel of vehicles in obstacle-dense environments.
Download

Paper Nr:	56
Title:	High-Precision Contour Tracking for Mobile Manipulators in Large-Scale Industrial Applications
Authors:	Buu Hai Dang Trinh, Daniel Heß and Christof Röhrig
Abstract:	Industrial manipulators are limited in their workspace due to mechanical constraints, which pose significant challenges in large-scale industrial applications. Expanding a robot’s workspace often involves deploying additional stationary manipulators or integrating linear axes, both of which increase installation costs and system complexity without gaining much flexibility. A more effective and flexible solution is to integrate industrial manipulators onto mobile platforms. To support this research, the authors developed a mobile manipulator system consisting of a mobile platform driven by two Differential Drive Steering Units and an industrial robotic arm with six Degrees of Freedom (DoF). This configuration provides the system with nine DoF in its configuration space, substantially extending the workspace compared to conventional fixed-base manipulators. A trajectory control method is proposed to ensure smooth, low-vibration, and high-precision motion during operation. To enable accurate localization, a cost-effective method based on a 2D laser sensor and artificial landmarks is introduced. Furthermore, a high-precision contour tracking algorithm is developed to monitor the position of the end-effector relative to the workpiece. The proposed methods are validated through real-world experiments, demonstrating millimeter-level accuracy in both positioning and tracking.
Download

Paper Nr:	93
Title:	Δ-Y Transformations in Manipulator’s Stiffness Analysis
Authors:	Alexandr Klimchik and Anatol Pashkevich
Abstract:	The paper proposes a Δ-Y transformations technique for stiffness modelling of over-constrained manipulators with internal cross-linkages. It allows representing complex structures as a serial-parallel equivalent one that can be easily handled by the VJM-based method. To derive desired analytical expressions for the equivalent serial-parallel structure, the MSA-based stiffness modelling approach is employed first, which allows describing the stiffness response for both the Δ and Y structures operating with VJM-type stiffness matrices. Further, the desired relations between equivalent Δ-Y and Y-Δ stiffness matrices are obtained. The example of stiffness modelling of a non-rigid Gough-Stewart platform with multiple cross-linkages demonstrates the benefits of the proposed technique
Download

Paper Nr:	98
Title:	CSDF-by-SIREN: Learning Signed Distances in the Configuration Space Through Sinusoidal Representation Networks
Authors:	Christoforos Vlachos and Konstantinos Moustakas
Abstract:	Signed Distance Functions (SDFs) are used in many fields of research. In robotics, many common tasks, such as motion planning and collision avoidance use distance queries extensively and, as a result, SDFs have been integrated widely in such tasks, fulfilling even the tightest speed requirements. At the same time, the idea of the more natural representation of distances directly in the configuration space (C-space) has been gaining ground, resulting in many interesting publications in the last few years. In this work, we aim to define a C-space Signed Distance Function (CSDF) in a way that parallels other SDF definitions. Additionally, coupled with recent advancements in machine learning and neural representation of implicit functions, we attempt to create a neural approximation of the CSDF in a way that is fast and accurate. To validate our contributions, we construct an experiment environment to test the accuracy of our proposed workflow in an inverse kinematics contact test. Comparing these results to the performance of another published approach to the neural implicit representation of distances in the Configuration Space, we found that our method offers a considerable improvement, reducing the measured errors and increasing the success rate.
Download

Paper Nr:	106
Title:	Pressing Force Regulation in Robotic 3D Printing via CFD-Aided Nozzle Posture Control
Authors:	Shinichi Ishikawa, Ryo Yamada, Wakana Tsuru and Ryosuke Tasaki
Abstract:	In the 3D printing process of applying materials, inadequate extrusion pressure critically deteriorates the deposition quality in robotic 3D printing. Feedback-based motion control with sensing and AI technology has been studied to respond to uneven and flexible surfaces, but challenges remain in optimizing the application force according to the situation. In this study, we investigate numerically and experimentally how the nozzle can change its orientation during the printing motion in a way that reduces the extrusion force. Numerical calculations are performed to derive the relationship between the clearance between the nozzle and the base and the extrusion force for multiple nozzle orientation patterns. Based on the numerical results, the relationship between the operating quantity (clearance) and the control quantity (maximum pressure and line width) is expressed mathematically to predict the quality model based on numerical fluid dynamics. In a variable-thickness printing experiment using a robot arm, a convex shape was reproduced by robot motion control that continuously changed the nozzle orientation. The experimental results demonstrated that adjusting the nozzle orientation effectively maintained extrusion force, preventing a reduction of approximately 0.01 N, as verified through force sensor-based inspection.
Download

Paper Nr:	115
Title:	A Digital Twin Enabled Runtime Analysis and Mitigation for Autonomous Robots Under Uncertainties
Authors:	Jalil Boudjadar and Mirgita Frasheri
Abstract:	Autonomous mobile robots are increasingly deployed in various application domains, often operating in environments with uncertain conditions. Such robots rely on the state and performance assessments at runtime to autonomously control the robot functionality. However, uncertainty can significantly impact the robot sensors and actuators making it challenging to assess the robot state and quantify its performance reliably. This paper proposes a digital twin (DT) asset for the runtime estimation and validation of state and performance for a mobile autonomous robot "Turtlebot3" (TB3) operating under uncertainties, namely Lidar sensor obstruction and unknown floor friction and density. The proposed DT setup enables real-time state synthesis post-uncertainty, so that to estimate the performance and validate it using TeSSLa monitors, and compute mitigation actions. To maintain the robot autonomy, our DT intervenes only when an uncertainty is identified. The experimental results demonstrate that our DT enables to eliminate 70% of the related uncertainty while it mostly maintains the real-time synchronization with the physical TB3 robot operating a frequency of 0.2s.
Download

Paper Nr:	117
Title:	HOI-LCD: Leveraging Humans as Dynamic Landmarks Toward Thermal Loop Closing Even in Complete Darkness
Authors:	Tatsuro Sakai, Yanshuo Bai, Kanji Tanaka, Wuhao Xie, Jonathan Tay Yu Liang and Daiki Iwata
Abstract:	Visual SLAM (Simultaneous Localization and Mapping) is a foundational technology for autonomous navigation, enabling simultaneous localization and mapping in diverse indoor and outdoor environments. Among its components, loop closure plays a vital role in maintaining global map consistency by recognizing revisited locations and correcting accumulated localization errors. Conventional SLAM methods have primarily relied on RGB cameras, leveraging feature-based matching and graph optimization to achieve high-precision loop detection. Despite their success, these methods are inherently sensitive to illumination conditions and often fail under low-light or high-contrast scenes. Recently, thermal infrared cameras have gained attention as a robust alternative, particularly in dark or visually degraded environments. While various thermal-inertial SLAM approaches have been proposed, they still depend heavily on static structures and visual features, limiting their effectiveness in textureless or dynamic environments. To address this limitation, we propose a novel loop closure method that utilizes Human-Object Interaction (HOI) as dynamic-static composite landmarks in thermal imagery. Although humans are conventionally considered unsuitable as landmarks due to their motion, our approach overcomes this by introducing HOI feature points as landmarks. These feature points exhibit both a human attribute, characterized by stable detection across RGB and thermal domains via person tracking, and a static-object attribute, characterized by contact with visually consistent, semantically meaningful objects. This duality enables robust loop closure even in dynamic, low-texture, and dark environments, where traditional methods typically fail.
Download

Paper Nr:	119
Title:	Design Space Exploration and Performance Evaluation of a Multi-Chamber, Multi-Curvature Soft Actuator for Robotic Applications
Authors:	Ansari Usama and Asokan Thondiyath
Abstract:	Soft actuators are finding wide applications in robotics due to the compliance they offer in handling delicate objects. The design of soft actuators is challenging due to the fragile nature of the materials used and the difficulty of fabricating them. Also, soft actuators must be designed to achieve the desired bending performance that suits the application. This paper presents the design and analysis of a multi-chamber, multi-curvature soft actuator for robotic gripping applications. This design combines two different configurations to get the desired bending curvature of the actuator. Modeling of the actuator and analysis of the effect of various design parameters on the bending angle and the tip force are presented. Prototype fabrication and experimental results are also presented. The results confirm that it is possible to custom-design soft actuators to meet specific performance requirements through design synthesis.
Download

Paper Nr:	137
Title:	Towards Fully Onboard State Estimation and Trajectory Tracking for UAVs with Suspended Payloads
Authors:	Martin Jiroušek, Tomáš Báča and Martin Saska
Abstract:	This paper addresses the problem of tracking the position of a cable-suspended payload carried by an unmanned aerial vehicle, with a focus on real-world deployment and minimal hardware requirements. In contrast to many existing approaches that rely on motion-capture systems, additional onboard cameras, or instrumented payloads, we propose a framework that uses only standard onboard sensors—specifically, real-time kinematic global navigation satellite system measurements and data from the onboard inertial measurement unit—to estimate and control the payload’s position. The system models the full coupled dynamics of the aerial vehicle and payload, and integrates a linear Kalman filter for state estimation, a model predictive contouring control planner, and an incremental model predictive controller. The control architecture is designed to remain effective despite sensing limitations and estimation uncertainty. Extensive simulations demonstrate that the proposed system achieves performance comparable to control based on ground-truth measurements, with only minor degradation (< 6%). The system also shows strong robustness to variations in payload parameters. Field experiments further validate the framework, confirming its practical applicability and reliable performance in outdoor environments using only off-the-shelf aerial vehicle hardware.
Download

Paper Nr:	140
Title:	NAMOUnc: Navigation Among Movable Obstacles with Decision Making on Uncertainty Interval
Authors:	Kai Zhang, Eric Lucet, Julien Alexandre Dit Sandretto, Shoubin Chen and David Filliat
Abstract:	Navigation among movable obstacles (NAMO) is a critical task in robotics, often challenged by real-world uncertainties such as observation noise, model approximations, action failures, and partial observability. Existing solutions frequently assume ideal conditions, leading to suboptimal or risky decisions. This paper introduces NAMOUnc, a novel framework designed to address these uncertainties by integrating them into the decision-making process. We first estimate them and compare the corresponding time cost intervals for removing and bypassing obstacles, optimizing both the success rate and time efficiency, ensuring safer and more efficient navigation. We validate our method through extensive simulations and real-world experiments, demonstrating significant improvements over existing NAMO frameworks. More details can be found in our website: https://kai-zhang-er.github.io/namo-uncertainty/.
Download

Paper Nr:	151
Title:	A Web-Controlled, Modular 3D-Printed Exoskeleton for Upper Limb Stroke Recovery
Authors:	Crina Bărbieru and Isabela Drămnesc
Abstract:	Stroke survivors often experience partial or complete loss of hand function, significantly affecting their ability to perform everyday tasks. Current rehabilitation methods can be resource intensive and require significant human intervention. This paper aims to develop a portable, modular, 3D-printed robotic hand exoskeleton that provides targeted repetitive exercises designed to enhance motor recovery. This exoskeleton is controlled via a web application, which includes progress-tracking functionalities for both patients and physical therapists, enabling remote monitoring. Preliminary testing was conducted with one patient to evaluate the usability and efficacy of the device. Feedback was collected from a physical therapist to assess the feasibility of the exoskeleton. The proposed system offers a scalable, cost-effective solution for post-stroke hand rehabilitation. Further studies with larger cohorts are needed to validate efficacy.
Download

Paper Nr:	152
Title:	Redundancy Resolution in Multiple Feasibility Maps via MultiFM-RRT
Authors:	Marc Fabregat-Jaén, Adrián Peidró, María Flores, Luis Payá and Óscar Reinoso
Abstract:	This paper presents MultiFM-RRT, a novel algorithm for redundancy resolution in kinematically redundant manipulators based on the exploration of multiple Feasibility Maps (FMs). They encode all valid configurations of redundant variables for a prescribed task trajectory, enabling global optimization and constraint satisfaction. Unlike traditional velocity-based methods, which are limited to local solutions, and grid-based methods, which are computationally intensive, MultiFM-RRT leverages the Rapidly-exploring Random Tree (RRT) framework to efficiently explore the space of feasible solutions across multiple feasibility maps. The algorithm incorporates singularity maps to enable transitions between different aspects, ensuring comprehensive coverage of the solution space. By computing feasibility maps online and guiding exploration with probabilistic sampling of goal and singularity sets, MultiFM-RRT achieves a balance between computational efficiency and global optimality. The proposed approach is demonstrated on a Stewart parallel platform, illustrating its ability to generate feasible, constraint-satisfying trajectories while efficiently handling redundancy.
Download

Paper Nr:	157
Title:	Place Recognition Using Bag of Semantic and Visual Words from Equirectangular Images
Authors:	María Flores, Marc Fabregat-Jaén, Juan José Cabrera, Adrián Peidró, David Valiente and Luis Payá
Abstract:	Place recognition has a crucial relevance in some tasks of mobile robot navigation. For example, it is used for the detection of loop-closure or for estimating the position of a mobile robot along a route in a known environment. If place recognition is based on visual information, it can be approached as an image retrieval problem. The Bag of Visual Words technique can be used for image retrieval. Image retrieval is based on an image representation (for example, a vector) that contains relevant visual information. In this paper, two image signatures are proposed. Both are based on semantic and visual information. A bag of visual words is created for each semantic class. Local feature descriptors are classified according to the projection of their associated point on a segmented semantic map. On the one hand, the image signature is composed of a set of histograms where each cell encodes the frequency with which a visual word appears in the image. On the other hand, the image signature is composed of a set of vectors where each cell encodes the sum of the cosine distance between the visual word and the nearest extracted features.
Download

Paper Nr:	162
Title:	Method for Automated Forklift Pallet Transfer with Simple Camera Calibration
Authors:	Tibor Bataljak Savić, Krešimir Turković, Ian Petek and Damjan Miklić
Abstract:	This paper describes a practical method for calibrating the camera pose for adaptive pallet pickup by automated forklifts. Adaptive pickup is an important prerequisite for human-robot collaborative workflows. It enables robots to handle pallets that have been incorrectly placed by humans. We propose a vision-based method that estimates the pallet pose from RGB-D data and adapts the robot approach path accordingly. The vision pipeline combines semantic segmentation of the RGB image with geometric analysis of the depth channel. Precise camera pose calibration is fundamental for the accuracy of the whole pipeline. The method relies on the known geometry of the forks and can be run on-line before every operation. This is important from a practical point of view, as it compensates for small deviations that may occur due to vibrations during vehicle motion. We present validation results in a simulated environment and on a real automated forklift.
Download

Short Papers

Paper Nr:	27
Title:	ORB-Based Map Fusion with Position Transformation for Enhanced Pairwise Connection
Authors:	Lucas Alexandre Zick, Dieisson Martinelli, Andre Schneider de Oliveira and Vivian Cremer Kalempa
Abstract:	This paper discusses developing and evaluating a map fusion algorithm based on ORB (Oriented FAST and Rotated BRIEF) feature matching, designed to improve the integration of robotic occupancy grids. The algorithm effectively merges maps generated by multiple robots, accommodating map size and orientation variations. A key aspect of its functionality is the ability to accurately position robots within the fused map, even when the overlap between maps is minimal. Comprehensive testing demonstrated the algorithm’s effectiveness in identifying correlations between different map pairs and aligning them accurately, as well as its capability to assess the success of the merging process, distinguishing between successful merges and those with inaccuracies. The findings indicate that this approach significantly enhances the capabilities of multi-robot systems, improving navigation and operational efficiency in complex environments.
Download

Paper Nr:	29
Title:	A Scalable Robot-Agnostic Voice Control Framework for Multi-Robot Systems
Authors:	Valentina Pericu, Federico Rollo and Navvab Kashiri
Abstract:	In recent years, Multi-Robot Systems (MRS) have gained increasing relevance in domains such as industrial automation, healthcare, and disaster response, offering effective solutions to manage complex and dynamic tasks. However, controlling such systems remains a challenge, particularly for users without expertise in robotics. A critical factor in addressing this challenge is developing intuitive and accessible Human-Robot Interaction (HRI) mechanisms that enable seamless communication between humans and robots. This paper introduces a scalable, robot-agnostic voice control framework designed to simplify interaction with MRS. The framework enables users to issue voice commands that are processed into actionable, robot-specific instructions through a centralized architecture. At its core, the framework features a centralized Control Management System (CMS) that is responsible for processing voice commands and interpreting them into robot-agnostic actions. System scalability is achieved through namespace management and a flexible structure, allowing new robots to be integrated and larger teams to be accommodated with minimal effort. By minimizing hardware requirements and leveraging voice commands as the primary interaction modality, the framework reduces technical barriers and provides an accessible, cost-effective solution for non-expert users. Experimental validation demonstrates its flexibility, scalability, and effectiveness in multi-robot scenarios. This work contributes to advancing HRI by offering a robust, intuitive, and adaptable solution for managing heterogeneous robot teams across dynamic environments.
Download

Paper Nr:	32
Title:	CLIP-LLM: A Framework for Autonomous Plant Disease Management in Greenhouse
Authors:	Muhammad Salman, Muhayy Ud Din and Irfan Hussain
Abstract:	Agricultural disease detection and intervention remain challenging due to complex crop health variations, dynamic environmental conditions, and labor-intensive fieldwork. We introduce an end-to-end, platformagnostic robotic pipeline for autonomous disease detection and treatment systems, with a specific focus on cassava leaves as an example. The pipeline integrates a vision-language perception module based on a pretrained Contrastive Language-Image Pre-training (CLIP) model, fine-tuned on an augmented dataset of cassava leaf images for disease detection. High-level task planning is performed by a Generative Pre-trained Transformer 4 (GPT-4), which interprets perception outputs and generates symbolic action plans (e.g., navigate to target, perform treatment). The low-level control system is implemented in the PyBullet dynamic simulator. We evaluated a vision-language model (VLM) perception and a Large Language Model (LLM) based planning system (in a virtual environment with predefined 3D coordinates for plant and spray positions). The VLM achieved 83% classification accuracy in simulation and real-time tests with a static camera produced classification accuracies of 70% Cassava Brown Streak Disease (CBSD), 65% Cassava Mosaic Disease (CMD) and 52% Cassava Bacterial Blight (CBB), and under dynamic camera it achieve the accuracy of 65% (CBSD), 52% (CMD), and 32% (CBB). Currently, our low-level controller executes the LLM-generated plans with high precision (less than ±2 mm positioning error). These results demonstrate the viability of our platform-agnostic modular architecture for precision agriculture that supports closed-loop robustness and scalability.
Download

Paper Nr:	42
Title:	Point Cloud Registration for Visual Geo-Referenced Localization Between Aerial and Ground Robots
Authors:	Gonzalo Garcia and Azim Eskandarian
Abstract:	Cooperative perception between aerial and ground robots relies on the accurate alignment of spatial data collected from different platforms, often operating under diverse viewpoints and sensor constraints. In this work, point cloud registration techniques for monocular visual SLAM-generated maps are investigated, which are common in lightweight autonomous systems due to their low cost and sensor simplicity. However, monocular visual SLAM outputs are typically sparse and suffer from scale ambiguity, posing significant challenges for map fusion. We evaluate registration pipelines combining coarse global feature matching with local refinement methods, including point-to-plane and plane-to-plane Iterative Closest Point alignments, to address these issues. Our approach emphasizes robustness to differences in scale, density, and perspective. Additionally, we assess the consistency of the resulting estimated trajectories to support geo-referenced localization across platforms. Experimental results using datasets from both aerial and ground robots demonstrate that the proposed methods improve spatial coherence by a factor of over 4 based on statistical metrics, and enable collaborative mapping and localization in GNSS-intermittent environments. This work can contribute to advancing multi-robot coordination for real-world tasks such as infrastructure inspection, exploration, and disaster response.
Download

Paper Nr:	55
Title:	Development of a Control System for an Innovative Parallel Robot Used in Laparoscopic Pancreatic Surgery
Authors:	Doina Pisla, Andra Ciocan, Bogdan Gherman, Diana Schlanger, Alexandru Pusca, Nadim Al Hajjar, Emil Mois, Andrei Cailean, Nicoleta Pop, Calin Vaida, Paul Tucan and Ionut Zima
Abstract:	This paper presents the development of the control architecture for an innovative parallel robot, designed to assist surgeons during the minimally invasive pancreatic cancer surgery. Based on the defined medical protocol and surgeon requirements. The robot was designed to serve as a surgical assistant and to manipulate a third active instrument. The system features a 3-DOF parallel active module coupled to a passive spherical module guiding the instrument through a Remote Center of Motion (RCM). The master-slave control architecture enables surgeons to operate the robot using a 3D Space Mouse or haptic device (Omega.7). The system automatically calculates RCM position using IMU sensors, validated through optical tracking.
Download

Paper Nr:	72
Title:	Predicting Contact Surfaces in Repetitive Robotic Tasks
Authors:	Luis Hernán Campos, José Luis Reyes Ramos, Marcelo Fajardo-Pruna, Christian Tutivén and Carlos Saldarriaga
Abstract:	Modern industrial robotics increasingly demands adaptive interaction with diverse materials in repetitive tasks, where traditional model-based control struggles to accommodate surface variability. This study introduces a novel framework that integrates impedance control with a machine learning-based surface classification system to enhance robotic adaptability in contact-rich environments. Using a 7-DOF Franka Emika Panda manipulator, we simulated repetitive trajectories over six material types and collected comprehensive dynamic interaction data. A CatBoostClassifier was trained on this dataset to predict surface type based on features such as joint torques, contact forces, and end-effector kinematics. The classifier achieved an overall accuracy of 99%, with F1-scores exceeding 0.98 across all classes, demonstrating robust discrimination, even between materials with similar frictional properties like brass and Teflon. Results show that our approach reduces manual tuning effort and maintains stable impedance responses under perturbations up to 50 N. This fusion of data-driven inference and classical control lays the groundwork for real-time parameter adaptation in robotic systems, offering new pathways toward safer, more efficient operation in unstructured industrial settings. Future work will expand the framework with multimodal sensing and evaluate its generalization on novel surfaces in physical deployments.
Download

Paper Nr:	77
Title:	Simultaneous Learning of State-to-State Minimum-Time Planning and Control
Authors:	Swati Dantu, Robert Pěnička and Martin Saska
Abstract:	This paper tackles the challenge of learning a generalizable minimum-time flight policy for UAVs, capable of navigating between arbitrary start and goal states while balancing agile flight and stable hovering. Traditional approaches, particularly in autonomous drone racing, achieve impressive speeds and agility but are constrained to predefined track layouts, limiting real-world applicability. To address this, we propose a reinforcement learning-based framework that simultaneously learns state-to-state minimum-time planning and control and generalizes to arbitrary state-to-state flights. Our approach leverages Point Mass Model (PMM) trajectories as proxy rewards to approximate the true optimal flight objective and employs curriculum learning to scale the training process efficiently and to achieve generalization. We validate our method through simulation experiments, comparing it against Nonlinear Model Predictive Control (NMPC) tracking PMM-generated trajectories and conducting ablation studies to assess the impact of curriculum learning. Finally, real-world experiments confirm the robustness of our learned policy in outdoor environments, demonstrating its ability to generalize and operate on a small ARM-based single-board computer.
Download

Paper Nr:	79
Title:	Real-Time Hand Gesture Control of a Robotic Arm with Programmable Motion Memory
Authors:	Daniel Giraldi Michels, Davi Giraldi Michels, Lucas Alexandre Zick, Dieisson Martinelli, André Schneider de Oliveira and Vivian Cremer Kalempa
Abstract:	The programming of industrial robots is traditionally a complex task that requires specialized knowledge, limiting the flexibility and adoption of automation in various sectors. This paper presents the development and validation of a programming by demonstration system to simplify this process, allowing an operator to intuitively teach a task to a robotic arm. The methodology employs the MediaPipe library for real-time hand gesture tracking, using a conventional camera to translate human movements into a robot-executable trajectory. The system is designed to learn a manipulation task, such as ’pick and place’, and store it for autonomous reproduction. The experimental validation, conducted through 50 consecutive cycles of the task, demonstrated the high robustness and effectiveness of the approach, achieving a 98% success rate. Additionally, the results confirmed the excellent precision and repeatability of the method, evidenced by a standard deviation of only 0.0126 seconds in the cycle time. A video demonstrating the system’s functionality is available for illustrative purposes, separate from the quantitative validation data. It is concluded that the proposed approach is a viable and effective solution for bridging the gap between human intention and robotic execution, contributing to the democratization of automation by offering a more intuitive, accessible, and flexible programming method.
Download

Paper Nr:	81
Title:	A New Sliding Mode Control Proposal with a Clegg Integrator for a Mobile Manipulator
Authors:	Pablo Proaño, Paulo Leica and Gabriela Andaluz
Abstract:	This article presents a novel control strategy for trajectory tracking in mobile manipulators. The proposed method combines a conventional Sliding Mode Controller (SMC) with a reset-based integrator, specifically a Clegg integrator, applied to the discontinuous component of the sliding surface. The system under study consists of a mobile platform with dynamic behavior and a robotic arm modeled kinematically. The main objective is to improve tracking performance and reduce control signal oscillations, particularly under abrupt reference changes and external disturbances. A reference trajectory with an inclined square shape is used to challenge the controller with sudden directional transitions. To evaluate the effectiveness of the proposed approach, both the classical SMC and the SMC+Clegg controllers are implemented and tested under the same conditions. The performance is analyzed using standard indices such as Integral Square Error (ISE), Integral Absolute Error (IAE), and Total Variation of the control signal (TVu). Results show that the proposed controller achieves improved trajectory tracking with reduced overshoot and chattering, while maintaining robustness to disturbances. Stability is formally demonstrated using Lyapunov theory. The positive impact of the Clegg integrator is highlighted in the discontinuous control component, allowing for reduced control effort without compromising tracking quality or disturbance rejection.
Download

Paper Nr:	82
Title:	Optimizing Collision Avoidance in Dynamic Multi-Robot Systems: A Velocity Obstacle and BB-PSO Approach with Priority Consideration
Authors:	Luis H. Sanchez-Vaca, Gildardo Sanchez-Ante and Hernan Abaunza
Abstract:	This study proposes integrating Reciprocal Velocity Obstacles (RVO) with Bare Bones Particle Swarm Optimization (BB-PSO) for prioritized motion planning in multi-robot systems. BB-PSO was chosen because it has fewer parameters to tune, reduced computational complexity, and provides potentially faster convergence compared to standard PSO. The methodology enables collision avoidance and path planning while allowing differentiated robot behaviors based on priority levels. Simulations used a two-phase experimental strategy: first, tuning cost function parameters through grid search, and second, evaluating various priority configurations and random scenarios. Results show that the selected weight configuration (α = 4,β = 2) balances goal-seeking and obstacle avoidance, enabling high-priority agents to move directly while ensuring overall group safety. Scenarios with higher average priorities exhibited shorter travel distances and faster completion times, whereas those with lower or imbalanced priorities led to more conservative behavior and delays. Compared to a greedy baseline, the proposed method significantly reduced collisions, achieving an average of 1.0 collision per scenario versus 6.6 with the greedy approach. Some priority configurations achieved complete task fulfillment without any collisions, highlighting the potential for optimized multi-robot coordination. The proposed method offers a promising strategy for prioritized motion planning, balancing efficiency and safety based on task importance. Future research includes comparing BB-PSO with other optimization methods, reducing sample requirements, dynamically adjusting priorities, and extending the model to incorporate task parameterizations and autonomous priority adaptation.
Download

Paper Nr:	84
Title:	Calibration Architecture for the Nonlinear Wheel Odometry Model with Integrated Noise Compensation
Authors:	Máté Fazekas and Péter Gáspár
Abstract:	In the motion estimation of self-driving vehicles, the three main requirements are accuracy, robustness, and cost-effectiveness. The generally applied sensors and methods are the GNSS, inertial, and visual-odometry, but the contradictory requirements demand the integration of new ideas. The wheel odometry could be an adequate choice since the method is robust and cost-effective, but the accuracy of the estimation is limited by the parameter uncertainty, thus a calibration method should be included as well. However, the general parameter identification of a nonlinear model in the presence of noise has not been solved yet. The presented method is based on the assumption that noisy, but several measurements of GNSS and IMU sensors are available in a self-driving vehicle. In the proposed architecture, nonlinear least squares and optimal control techniques are combined in a unique way to compensate for the noise of the orientation and wheel rotation signals to achieve unbiased model calibration. The performance of the developed algorithm and the accuracy of parameter estimation are demonstrated with detailed validation and a test with a real vehicle.
Download

Paper Nr:	99
Title:	3D Hand-Eye Calibration for Collaborative Robot Arm: Look at Robot Base Once
Authors:	Leihui Li, Lixuepiao Wan, Volker Krueger and Xuping Zhang
Abstract:	Hand-eye calibration is a common problem in the field of collaborative robotics, involving the determination of the transformation matrix between the visual sensor and the robot flange to enable vision-based robotic tasks. However, this process typically requires multiple movements of the robot arm and an external calibration object, making it both time-consuming and inconvenient, especially in scenarios where frequent recalibration is necessary. In this work, we extend our previous method which eliminates the need for external calibration objects such as a chessboard. We propose a generic dataset generation approach for point cloud registration, focusing on aligning the robot base point cloud with the scanned data. Furthermore, a more detailed simulation study is conducted involving several different collaborative robot arms, followed by real-world experiments in an industrial setting. Our improved method is simulated and evaluated using a total of 14 robotic arms from 9 different brands, including KUKA, Universal Robots, UFACTORY, and Franka Emika, all of which are widely used in the field of collaborative robotics. Physical experiments demonstrate that our extended approach achieves performance comparable to existing commercial hand-eye calibration solutions, while completing the entire calibration procedure in just a few seconds.
Download

Paper Nr:	109
Title:	Intelligent Surveillance System Using Deep Learning to Reduce Shoplifting in Minimarkets in Santiago de Surco, Lima, Peru
Authors:	Yosep Alexeis Solorzano Aguero and Jose Karim Candela Rengifo
Abstract:	This article presents an intelligent video surveillance system for theft detection in minimarkets located in Santiago de Surco, Lima. The proposed solution integrates computer vision techniques with deep learning models such as Convolutional Neural Networks (CNN) and You Only Look Once (YOLO), implemented using PyTorch. The system analyzes customer movements in real time to detect suspicious behavior patterns, including torso twists and concealment attempts. Trained on a dataset of over 2700 real and simulated images, the model achieved an accuracy of 82%, outperforming traditional surveillance systems by more than 30%. The solution includes a web interface developed with FastAPI (Fast Application Programming Interface, a high-performance Python framework for building APIs) and Angular, enabling remote monitoring. Practically, the system can reduce economic losses by up to 15%, offering a scalable and cost-effective alternative for improving security in small commercial environments.
Download

Paper Nr:	112
Title:	Manipulability Maximization of a Liquid-Handling Manipulator for Sloshing Suppression via Container Tilting
Authors:	Ryuji Nakagawa and Ryosuke Tasaki
Abstract:	The manipulability index measures a robot’s motion capability. To avoid singularity problems and to achieve unexpected changes in tasks, a method that explicitly considers this metric is needed. However, the index is a nonlinear function that depends on the state of the manipulator, making optimization difficult in a short period of time. In addition, real-time control of the robot, which requires computational efficiency, is necessary to ensure safety in a dynamic environment. Previous studies have generated trajectories that increase this metric, but different tasks require different constraints to be considered. Controlling not only the trajectory but also the posture and velocity of the end-effector expands the area of practical use. In this paper, we formulate a manipulability optimization problem for real-time control in liquid transfer and solve it efficiently using inequality constraints. In liquid transfer experiments, the method successfully generates safe and kinematic trajectories with high performance by optimizing both manipulability and controlling velocity and attitude to suppress liquid surface vibration.
Download

Paper Nr:	136
Title:	Robust Scene Understanding for Mobile Robots Based on Vision and Deep Learning Models
Authors:	Leticia C. Pereira and Fernando S. Osório
Abstract:	This paper presents the architecture and results of AIVFusion, a real-time perception system designed to generate a rich, multi-layered understanding of an environment from a single monocular camera for autonomous mobile robots. The system is designed to fuse information from different deep learning models to achieve a comprehensive scene understanding. Our architecture integrates three open-source models to perform distinct perception tasks: object detection (YOLOv8), semantic segmentation (FastSAM), and monocular depth estimation (Depth Anything V2). By fusing these outputs, the system generates a unified representation that identifies the navigable area, detects nearby obstacles based on depth information, and semantically labels those identified as “person”. The resulting perceptual information can then be leveraged by higher-level systems for tasks such as decision-making and safer navigation. The system’s viability is demonstrated through qualitative tests in indoor environments. These results confirm its ability to operate in real-time (approximately 10 FPS) and to effectively fuse the perception layers, even in challenging scenarios involving partial object occlusion.
Download

Paper Nr:	147
Title:	Augmenting Neural Networks-Based Model Approximators in Robotic Force-Tracking Tasks
Authors:	Kevin Saad, Vincenzo Petrone, Enrico Ferrentino, Pasquale Chiacchio, Francesco Braghin and Loris Roveda
Abstract:	As robotics gains popularity, interaction control becomes crucial for ensuring force tracking in manipulator-based tasks. Typically, traditional interaction controllers either require extensive tuning, or demand expert knowledge of the environment, which is often impractical in real-world applications. This work proposes a novel control strategy leveraging Neural Networks (NNs) to enhance the force-tracking behavior of a Direct Force Controller (DFC). Unlike similar previous approaches, it accounts for the manipulator’s tangential velocity, a critical factor in force exertion, especially during fast motions. The method employs an ensemble of feedforward NNs to predict contact forces, then exploits the prediction to solve an optimization problem and generate an optimal residual action, which is added to the DFC output and applied to an impedance controller. The proposed Velocity-augmented Artificial intelligence Interaction Controller for Ambiguous Models (VAICAM) is validated in the Gazebo simulator on a Franka Emika Panda robot. Against a vast set of trajectories, VAICAM achieves superior performance compared to two baseline controllers.
Download

Paper Nr:	158
Title:	Fire-Resistant Wall-Climbing UAV for Victim Detection in Urban Search and Rescue Missions
Authors:	Vedant Hambire, Harsh Yadav, Dipshikha Hazari and Satyam Singh
Abstract:	Unmanned Aerial Vehicles (UAVs) have become invaluable in high-stakes search and rescue operations in fire-prone and fire-damaged environments due to their capabilities in victim locating and situation analysis. This paper describes the design, simulation, and realization of a fire-resistant, wall-climbing UAV with a human alive detection system powered by AI. The UAV includes a custom-designed H-frame made out of PLA which is thrust vectoring EDFs attached to a tilt-rotor system permitting vertical hovering and traversing. Structural and aerodynamic aspects were verified with FEA and CFD simulations performed on SimScale. To allow for autonomous victim detection, the UAV system includes a real-time human detector based on YOLOv8 with and optical flow and MediaPipe-based eye tracking to classify people as conscious, unconscious, dead, or blocked out. The UAV's mission computer, which comprises a Raspberry Pi with ROS, records would-detect status and location, and outputs tagged geo-coordinates for mission planning in real-time. Simulation and ground testing would confirm the system’s viability in heat-intensive, debris-laden environments, advancing the development of autonomous aerial platforms for disaster response, firefighting, and urban search and rescue (USAR) operations.
Download

Paper Nr:	45
Title:	Design and Control of a New Wrist Rehabilitation Robot
Authors:	Simona-Daiana Stiole, Pusca Alexandru, Paul Tucan, Iuliu Nadas, Vasile Bulbucan, Andrei Cailean, Dragos Sebeni, Alexandru Banica, Daniela Jucan, Radu Morariu, Calin Vaida, Petru Dobra, Jose Machado and Doina Pisla
Abstract:	This paper presents the design and control of a cost-effective wrist rehabilitation robot with the aim of providing an accessible and scalable solution for patients in need of upper-limb motor recovery. The primary goal is to create a compact system that can support repetitive and controlled wrist movements, particularly for individuals recovering from stroke. The robot’s mechanical structure, forward kinematic model and dynamic model were defined to minimize cost without compromising essential therapeutic functionality. Three control strategies were implemented and evaluated in simulation, including Independent Joint Control, Linear Quadratic Regulator, and an observer-based version using a Luenberger estimator for situations where only position sensors are available. These simulations serve to assess the feasibility of each control method in terms of performance, complexity, and compatibility with low-cost components for future hardware development.
Download

Paper Nr:	52
Title:	FOPID-Based Trajectory Control for an Unmanned Aerial Robotic Manipulator
Authors:	Gabriela M. Andaluz, Zahid Nazate, Paulo Leica and Guillermo Palacios-Navarro
Abstract:	This work presents a Fractional Order PID (FOPID) control strategy for trajectory tracking of an Unmanned Aerial Manipulator (UAM), proposed as an alternative to the conventional PID controller. Unlike classical integer-order controllers, the FOPID design enables more flexible tuning of the aerial manipulator’s kinematic response by introducing five independent tuning parameters. This added flexibility enhances system stability and improves robustness against abrupt reference changes. The controller parameters are optimized through Integral of Squared Error (ISE) minimization to ensure efficient performance. Simulation results confirm that the FOPID controller achieves superior trajectory tracking accuracy compared to the conventional PID. Specifically, the ISE values obtained with the FOPID reflect reductions of 23.46%, 24.99%, and 15.35% in the tracking errors along the 𝑥෤, 𝑦෤ and 𝑧̃ directions, respectively. These results validate the effectiveness of the FOPID approach in improving the control performance of unmanned aerial manipulators.
Download

Paper Nr:	53
Title:	Towards Universal Detection and Localization of Mating Parts in Robotics
Authors:	Stefan Marx, Attique Bashir and Rainer Müller
Abstract:	A key task in robotics is the precise joining of two components. This approach focuses on detecting basic geometric shapes such as rectangles, triangles, and circles, etc. on the respective mating counterparts. This paper first examines how precisely individual geometric shapes can be localized using stereoscopy with a single camera on the robot arm. After the localization of the individual shapes, the spatial relationships between these shapes are analyzed and then compared with those of a possible joining partner. If several features match, transformation parameters are calculated to define the optimal alignment for an accurate and efficient assembly. This method emphasizes simplicity and effectiveness in identifying complementary geometries for precise positioning during assembly tasks.
Download

Paper Nr:	59
Title:	Model Predictive Control for Crowd Navigation via Learning-Based Trajectory Prediction
Authors:	Mohamed Parvez Aslam, Bojan Derajic, Mohamed-Khalil Bouzidi, Sebastian Bernhard and Jan Oliver Ringert
Abstract:	Safe navigation in pedestrian-rich environments remains a key challenge for autonomous robots. This work evaluates the integration of a deep learning-based Social-Implicit (SI) pedestrian trajectory predictor within a Model Predictive Control (MPC) framework on the physical Continental Corriere robot. Tested across varied pedestrian densities, the SI-MPC system is compared to a traditional Constant Velocity (CV) model in both open-loop prediction and closed-loop navigation. Results show that SI improves trajectory prediction-reducing errors by up to 76% in low-density settings-and enhances safety and motion smoothness in crowded scenes. Moreover, real-world deployment reveals discrepancies between open-loop metrics and closed-loop performance, as the SI model yields broader, more cautious predictions. These findings emphasize the importance of system-level evaluation and highlight the SI-MPC framework’s promise for safer, more adaptive navigation in dynamic, human-populated environments.
Download

Paper Nr:	60
Title:	Comparative Analysis of Robotic Topologies for Transmission Line Inspection
Authors:	Davi Riiti Goto Valle, Ronnier Frates Rohrich and André Schneider de Oliveira
Abstract:	Power transmission line inspection plays a crucial role in maintaining the integrity and reliability of electrical infrastructure. With the increasing complexity of transmission line systems, robotic systems have emerged as a viable solution to automate the inspection process. This paper presents an analysis of three distinct robotic platforms designed for transmission line inspection. Each robot employs different topologies and mechanisms to perform the task, which are simulated environment. The paper compares the design, functionality, and simulation results of each robot, highlighting their strengths, weaknesses, and potential for real-world application.
Download

Paper Nr:	70
Title:	ROSBLOCKS: A Visual Programming Interface for ROS2 Robots
Authors:	Fernando Costa Nogueira, Dieisson Martinelli, Lucas Alexandre Zick, André Schneider de Oliveira and Vivian Cremer Kalempa
Abstract:	This work presents the development of a visual programming interface for robots compatible with ROS2, called ROSBLOCKS, using a modern architecture based on React, Blockly, Node.js, and Electron. The proposal aims to make robot programming more accessible, especially in educational contexts, by allowing users to create complex robotic behaviors through visual blocks, without the need for prior knowledge in programming languages such as Python or C++. The system is cross-platform and flexible, working with both simulated and physical robots that use ROS2, and allows for automatic code generation and execution from the visual assembly. Additionally, the system was designed to facilitate integration with different types of ROS2 topics, services, and actions. The system was tested in the classroom with undergraduate students who already have practical experience with ROS, enabling an assessment of its applicability in real teaching scenarios and allowing the observation of gains in productivity, engagement, and clarity in the construction of robotic behaviors. The complete source code and all validation materials from this study are openly available on GitHub at https://github.com/ferssor/rosblocks.
Download

Paper Nr:	83
Title:	Curvature-Constrained Motion Planning and Control for Traffic Cone Manipulation Robot
Authors:	Rudolf Krecht and Áron Ballagi
Abstract:	This paper presents an integrated system for traffic cone manipulation using a heavy-duty mobile robot equipped with GNSS-RTK localization, a custom remote supervision and mission control interface, and a curvature-constrained motion controller. Designed for use in semi-structured outdoor environments, the robot receives waypoint and speed commands via a tailored extension of Foxglove Studio, which enables intuitive map-based interaction and real-time trajectory editing. Owning to its high payload capacity, the platform prioritizes stability over maneuverability, thus, it cannot change orientation without longitudinal movement. To address this, we propose a smooth, curvature-based controller that enforces a minimum turning radius while following pose and heading goals. The system architecture is built on Robot Operating System 2 (ROS 2), leveraging modular nodes for map visualization, path planning, motion execution, and action triggering. Our experiments demonstrate the system’s ability to navigate complex waypoint paths and pause precisely at mission-dictated locations, more specifically cone placement locations. Our results show that even under turning constraints, the robot reliably executes full cone manipulation routines with high spatial accuracy and operational safety. The system highlights the feasibility of pairing high-level operator interfaces with low-level kinematic-aware planning for constrained robotic platforms.
Download

Paper Nr:	121
Title:	Simulation-Driven Design and Optimization of a Parametric Flat-Foot with Elastic Pads for a Planar Biped Robot
Authors:	Koray Kadir Şafak and Oğuzhan Aykut Ekşioğlu
Abstract:	This paper presents the simulation-driven design and optimization of a compliant foot for a planar biped robot. To enhance walking stability and reduce joint torques, 3D-printed elastic pads were fabricated and experimentally characterized through compression testing. These prototypes provided baseline stiffness and damping ranges that served as inputs to the simulation model. Using these data as a starting point, a genetic algorithm optimized pad parameters to minimize joint torque overloads while maintaining gait stability. Walking simulations were performed in MATLAB Simulink on flat terrain, comparing a rigid flat-foot with the optimized compliant foot with pads. Results demonstrated up to 46% reduction in peak hip torques and 35% reduction in knee torques, along with smoother contact forces and stable zero moment point (ZMP) trajectories. The study confirms that introducing passive compliance at the foot level improves bipedal locomotion efficiency without additional actuation.
Download

Paper Nr:	122
Title:	Design and Implementation of a Robotized Laser Module for Weed Control
Authors:	Kai Blanco, Luis Emmi and Roemi Fernández
Abstract:	This paper presents the design of a low-cost, modular system mounted on a mobile platform for weed control using laser technology. This proposal seeks to find an effective and sustainable solution for selective weed management in agricultural settings, avoiding harmful methods such as herbicides. The methodology for this work was based on the application of divergent-convergent thinking stages. Additionally, studies were conducted on potential movement systems, and in line with the system’s needs, a Core XY movement was selected. Standard elements and custom designs were adapted to the previous structure. Similarly, an analysis of potential casing designs was carried out, and through a convergent process, a design suitable for its function was selected. The results obtained in this work, such as the estimated movement system accuracy of less than 0.2mm and the simulated treatment time of 3.62 seconds, in an estimated area of 0,25 m2, demonstrate the feasibility of creating an effective, small-sized, and low-cost weed control system for users, providing the necessary precision to avoid damage to surrounding crops.
Download

Paper Nr:	131
Title:	Experimental Validation of Load Attitude Estimation Using Computer Vision and IMU-Based Approaches for Slung-Load Aerial Robots
Authors:	Shlok Panchal, Barbie Sharma, Yash Dadheech, Darshil Shah, Ayush Agnihotri, Kalash Jain, Parth S. Thakar and Anilkumar Markana
Abstract:	With the rise of the drone industry, there has been a surge in demand for its applications. One such critical application is using drones to transport suspended cargo, which requires minimal swing of the load. To achieve this, designing a robust control strategy plays a vital role. Such systems while in operation have critical issue of maintaining stability due to the interacting multi-body dynamics. Furthermore, a quadrotor with a slung load showcases coupled underactuated dynamics that complicates the control design problem. To effectively execute control implementation for such systems accurate feedback of load attitude becomes essential. For that matter, this study proposes two different approaches to determine the load attitude, namely, the computer vision (CV) based method using ArUco markers and the inertial measurement unit (IMU) based approach. The study investigates the real-time feasibility of these approaches through their response frequencies and tracking accuracies by comparing the experimental plots with their simulation counterpart, considering that as an ideal scenario. We also provide the implementation algorithms for both the methods proposed here. Finally, we conclude the findings by throwing light on their suitability to various slung load scenarios with variable swing angle ranges, also dwelling into the steady state behaviour comparisons in both the cases.
Download

Paper Nr:	133
Title:	Hematoxylin and Eosin Stained Histopathological Image Enhancement Method
Authors:	Bogusław Cyganek
Abstract:	Hematoxylin and eosin staining is one of the most well-known and common methods of staining histopathological samples. Its main purpose is to highlight the morphological features of tissues, which help doctors make the right diagnosis. However, it is not without its flaws, and the scans obtained in this way are characterized by high inconsistency not only resulting from the variability of the tissues themselves, but also due to the chemical reagents used, the technique of preparing the preparation, etc. This causes various difficulties and errors in the case of tissue assessment performed by the algorithm, but can also be a hindrance for doctors. Therefore, there are many methods to improve the quality of scans obtained from tissue stained in the H&E way. In this article, we present a fairly recent idea and very preliminary results for the use of our multi-channel virtual high-dynamic range MVHDR method to improve the parameters of H&E scans. Our method allows both data augmentation for CNN, but also significant detail enhancement that helps doctors identify the disease.
Download

Paper Nr:	149
Title:	Leveraging ROS to Support LLM-Based Human-Robot Interaction
Authors:	Walleed Khan, Deeksha Chandola, Enas AlTarawenah, Baran Parsai, Ishan Mangrota and Michael Jenkin
Abstract:	Large Language Model (LLM)-based systems have found wide application in providing an interface between complex systems and human users. It is thus not surprising to see interfaces between autonomous robots also adopting this strategy. Many modern robot systems utilize ROS as a middleware between hardware devices, standard software tools, and the higher level system requirements. Here we describe efforts to leverage LLM and ROS to provide not only this traditional middleware infrastructure but also to provide the audio- and text-based interface that users are beginning to expect from intelligent systems. A proof of concept implementation is described as well as an available set of tools to support the deployment of LLM-based interfaces to ROS-enabled robots and stationary interactive systems.
Download

Paper Nr:	154
Title:	Kinematic and Dynamic Analysis of Quadruped Legged Robots: A New Formulation Approach
Authors:	Vyshak Sureshkumar, Khalifa H. Harib and Adewale Oriyomi Oseni
Abstract:	This paper presents a framework for efficient kinematic and dynamic modelling of a quadruped robot using the recursive Newton-Euler method. The robot features 12 actuators-three per leg-and is analysed under both static walking and dynamic trotting gaits. The formulation incorporates assumed ground reaction forces and system over-constraints, enabling the resolution of contact forces through a reduced set of six linear equations. Twelve generalized coordinates are used for static gait analysis, with an additional generalized coordinate introduced for dynamic trotting. Body attitude, velocity, and acceleration are derived from joint-space trajectories, and forward dynamics is computed by inverting the inverse dynamics equations by numerically evaluating the mass matrix and nonlinear torque vectors. By employing a reduced set of generalized coordinates and simplified constraint handling of ground reactions, the proposed framework streamlines rigid-body dynamic simulation for such a high-degree-of-freedom system.
Download

Area 4 - Signal Processing, Sensors, Systems Modelling and Control

Full Papers

Paper Nr:	74
Title:	Filtering of Polytopic-Type Uncertain State-Delayed Noisy Systems
Authors:	Eli Gershon
Abstract:	The problem of H∞ state estimation is considered for uncertain polytopic retarded linear discrete-time stochastic systems. We first bring the solution of the estimation problem for the nominal case based on a previously developed BRL for state-delayed stochastic systems. We then extend our solution to the robust uncertain polytopic case where a vertex-dependent approach is applied. The latter is achieved via the application of a modified version of the Finsler lemma. The use of this lemma enable us to derive a solution which is less conservative comparing to the ”quadratic” solution where a single Lyapunov function is applied over all the uncertain polytope. The solution obtained for the robust case is composed of a set of LMIs based on only two tuning parameters. The theory presented is demonstrated by a numerical example.
Download

Short Papers

Paper Nr:	43
Title:	Space-Filling Regularization for Robust and Interpretable Nonlinear State Space Models
Authors:	Hermann Klein, Max Heinz Herkersdorf and Oliver Nelles
Abstract:	The state space dynamics representation is the most general approach for nonlinear systems and often chosen for system identification. During training, the state trajectory can deform significantly leading to poor data coverage of the state space. This can cause significant issues for space-oriented training algorithms which e.g. rely on grid structures, tree partitioning, or similar. Besides hindering training, significant state trajectory deformations also deteriorate interpretability and robustness properties. This paper proposes a new type of space-filling regularization that ensures a favorable data distribution in state space via introducing a data-distribution-based penalty. This method is demonstrated in local model network architectures where good interpretability is a major concern. The proposed approach integrates ideas from modeling and design of experiments for state space structures. This is why we present two regularization techniques for the data point distributions of the state trajectories for local affine state space models. Beyond that, we demonstrate the results on a widely known system identification benchmark.
Download

Paper Nr:	57
Title:	Wind Farm Power Prediction Using a Machine Learning Surrogate Model from a First-Principles Simulation Model
Authors:	Sebastian E. Pralong, Samuel Martínez-Gutiérrez, Dan E. Kröhling, Alejandro Merino, Gonzalo E. Alvarez, Daniel Sarabia and Ernesto C. Martínez
Abstract:	Reliable forecasting of wind farm power generation is essential for ensuring seamless grid integration and optimizing energy management strategies. This paper presents an integrated framework combining a first-principles simulation model of wind turbines as a data source for machine learning techniques to forecast wind farm power output. The simulation model accounts for wind speed, direction, temperature, and other climate variables, and is computationally intensive due to the need to account for the dynamics of each turbine operation, the wake effects, etc. To diminish the computational cost, this work introduces a surrogate Gaussian Processes (GPs) model that approximates the complex simulation model to provide predictions of both the mean and variance of power generation. To forecast future climate conditions, we employ a NARX (Nonlinear Autoregressive with Exogenous Inputs) neural network trained on historical data to account for wind speed, direction, and atmospheric conditions for the next two hours. The NARX model forecasts and the GPs predictions enable fast and accurate real-time forecasting of power generation for the entire wind farm. This approach significantly reduces computational times from hours to seconds while maintaining high accuracy, offering a scalable and efficient solution for real-time wind farm power prediction and online optimization.
Download

Paper Nr:	63
Title:	Experimental Evaluation of Camouflage Effectiveness Against Ground-Based Surveillance
Authors:	Viktor Vitoul, Jan Ivan, Ladislav Potužák, Michal Šustr and Barbora Hanková
Abstract:	Camouflaging mortar firing positions represents a critical force protection measure in modern conflicts, aiming to prevent enemy observation and subsequent destruction. The objective of this pilot study is to evaluate the effectiveness of various camouflage techniques in concealing mortars, ammunition assets, and support equipment from detection by selected ground-based reconnaissance means. The experimental phase employed a range of artillery reconnaissance sensors, optical devices, and unaided visual observation. The observed targets including mortar firing positions of various calibres and decoy positions were camouflaged using different methods and levels of concealment, and deployed in terrain with varying vegetation density and spatial characteristics. The detected differences in target visibility highlight the strengths and limitations of individual observation methods depending on target characteristics and environmental conditions. The findings of this pilot study offer practical recommendations for the effective camouflage of mortar units in current operational environments.
Download

Paper Nr:	75
Title:	Video-Based Vibration Analysis for Predictive Maintenance: A Motion Magnification and Random Forest Approach
Authors:	Walid Gomaa, Abdelrahman Wael Ammar, Ismael Abbo, Mohamed Galal Nassef, Tetsuji Ogawa and Mohab Hossam
Abstract:	Condition monitoring of high-speed machinery is critical to prevent unexpected breakdowns that could lead to injuries and cost billions. Traditional contact-based vibration sensors face limitations including measurement perturbations, point-specific data coverage, and installation constraints. This paper presents a novel non-contact machinery fault detection framework combining Eulerian video motion magnification with machine learning classification. The methodology comprises two integrated components. Primarily, a video-based vibration analysis pipeline utilizing Eulerian motion magnification with dense optical flow, which accomplish comprehensive signal processing for feature extraction using Fast Fourier transform. Then, a Random Forest classifier trained on video-derived temporal and frequency domain features. The system was validated based on ground-truth data from the Gunt PT500 machinery diagnosis and the Gunt TM170 balancing apparatus under four operational conditions: normal operation, outer ring bearing fault, and two imbalance severities (10g and 37g). Hence, experimental results demonstrate exceptional performance with 96.7% overall accuracy and a macro-averaged F1-score of 0.965 in discriminating fault conditions using solely video-derived features. The video processing allowed to identify distinct vibration signatures, from imbalance conditions showing amplitude variations to proportional fault severity ultimately offering a cost-effective solution for industrial condition monitoring applications.
Download

Paper Nr:	85
Title:	Towards Machine Learning Driven Virtual Sensors for Smart Water Infrastructure
Authors:	Vineeth Maruvada, Karamjit Kaur, Matt Selway and Markus Stumptner
Abstract:	Water utilities around the world are under increasing pressure from climate change, urban expansion, and aging infrastructure. To address these challenges, smarter and more sustainable water management solutions are essential. This study explores the use of Machine Learning (ML) to develop Virtual Sensors for smart water infrastructure. Virtual Sensors can complement or replace physical sensors while improving environmental sustainability and enabling reliable and cost-effective Digital Twins (DTs). Our experimental results show that several ML models outperform traditional methods such as Auto-Regressive Integrated Moving Average (ARIMA) in terms of forecast accuracy and timeliness. Among these, Extreme Gradient Boosting (XGBoost) and Long Short-Term Memory (LSTM) offer the best balance between accuracy and robustness. This research provides preliminary evidence that ML models can enable Virtual Sensors capable of delivering short-term forecasts. When successfully implemented, Virtual Sensors can transform water utilities by improving environmental sustainability, operational intelligence, adaptability, and resilience within Digital Twins.
Download

Paper Nr:	87
Title:	Unsupervised Analysis of Cyclist Performance for Route Segmentation and Ranking
Authors:	Rensso Mora-Colque and William Robson Schwartz
Abstract:	This paper presents a study on the analysis of cycling tours along a designated route, addressing the limited attention given to non-professional cyclists in existing research. Unlike previous work focused on elite athletes, this study considers a broader population, including commuters, recreational riders, and fitness-oriented cyclists. Data was collected using advanced sensors to capture diverse ride characteristics. An unsupervised learning approach was applied to segment cyclists based on behavioral and performance patterns. Furthermore, a novel ranking method based on genetic algorithms was developed to classify and prioritize cyclist groups meaningfully. Experiments were conducted on a newly proposed dataset tailored to this objective, enabling deeper insights into cycling dynamics across user types. The results validate the effectiveness of both the segmentation and ranking methods, offering practical implications for route planning and cyclist-focused infrastructure management.
Download

Paper Nr:	96
Title:	Sky Savers: Leveraging Drone Technology for Victim Localization in Avalanche Rescue via Transceiver Signal Analysis
Authors:	Robin Vetsch, Samuel Kranz, Tindaro Pittorino, Peter de Baets, Martial Châteauvieux, Christoph Würsch, Daniel Lenz and Sebastien Gros
Abstract:	In modern avalanche rescues, the search for buried victims is carried out primarily using a state-of-the-art handheld transceiver. However, in situations where the rescuers do not have the necessary experience, or if the victims are buried in areas that can be dangerous for the rescuers, e.g. due to the risk of secondary avalanches, this search process can be time-consuming, complex and dangerous. To overcome these challenges, we propose a proof-of-concept (PoC) of a search system based on an autonomous vertical take-off and landing (VTOL) aircraft that could significantly reduce search time, even in the case of multiple overlapping signal sources attributable to multiple victims or where conventional methods are not sufficiently efficient, e.g. in the case of large-scale avalanches. Electric drones or VTOL systems cannot be used because electro-magnetic interference (EMI) blocks the signal from the sending avalanche transceiver. By replacing electromagnetically noisy DC motors with a turbine, we effectively reduce electro-magnetic interference in the signal stream and demonstrate sub-meter localization accuracy under realistic field conditions. We employ a two-stage Extended Kalman Filter (EKF) approach to estimate the stationary target coordinates. Eventually, a VTOL system also allows for operations in adverse weather and rugged alpine terrain, greatly extending the practical capability of search and rescue missions.
Download

Paper Nr:	102
Title:	Optimizing Sensor Deployment Strategy for Tracking Mobile Heat Source Trajectory
Authors:	Thanh Phong Tran, Laetitia Perez, Laurent Autrique, Edouard Leclercq, Syrine Bouazza and Dimitri Lefevbre
Abstract:	Previous studies have investigated inverse problems in physical systems described by partial differential equations, particularly for identifying unknown parameters of mobile heat sources. An iterative minimization of a quadratic cost function, based on the conjugate gradient method, has shown reliable results in identifying heat densities and trajectories both offline and online. Although fixed sensor arrays can be effective, covering the full operating range of a moving heat source requires a large number of sensors, leading to inefficiencies and waste. A more efficient approach uses fewer mobile sensors mounted on autonomous robots. However, this introduces challenges in robot control, ensuring optimal positioning, coordination, and collision avoidance. To address this, we propose a method that combines sensitivity-based sensor placement with robot assignment algorithms such as the Hungarian Algorithm and Multi-Agent Path Finding. This enables effective tracking of the heat source’s trajectory while optimizing sensor deployment. The approach not only increases overall sensitivity of the sensor network but also improves identification performance with reduced latency and higher accuracy.
Download

Paper Nr:	150
Title:	On the Synthesis of Stable Switching Dynamics to Approximate Limit Cycles of Nonlinear Oscillators
Authors:	Nils Hanke, Zonglin Liu and Olaf Stursberg
Abstract:	This paper presents a novel method for approximating periodic behavior of nonlinear systems by use of switching affine dynamics. While previous work on approximating limit cycles by switching systems has been restricted to state space partitions with only two regions or approximations in the plane, this study employs more general partitions in higher-dimensional spaces as well as external signals to develop a scheme for synthesizing models with guaranteed existence of a globally stable limit cycle. The synthesis approach is formulated as a constrained numeric optimization problem, starting from sampled nonlinear dynamics data. It minimizes deviations between this data and the switching affine model’s limit cycle, while satisfying constraints to ensure global stability. The principle and effectiveness of the proposed method is illustrated through examples.
Download

Paper Nr:	160
Title:	High-Level Synthesis of an Efficient Hardware Implementation for a Smart Tactile Sensing System
Authors:	María-Luisa Pinto-Salamanca, Wilson-Javier Pérez-Holguín and José Antonio Hidalgo-López
Abstract:	Complex algorithms’ execution can be improved by means of carefully designed digital hardware. These take advantage of parallelization techniques, heterogeneous architectures, pipelines, and reuse of functional blocks, among others, to achieve low power consumption, low area overhead, and high real-time operating speeds. However, the design process for these architectures is typically long and complex compared to equivalent software implementations. This paper presents the design process and implementation of a hardware architecture designed to reconstruct triaxial contact forces for innovative tactile sensing systems. Such algorithms were implemented in hardware on a field-programmable gate array (FPGA) platform using a high-level hardware design approach, which allows for a significantly reduced design effort and accelerates the validation process of system functionality. The hardware design flow considers the integrated circuit design cycle and is based on the evaluation of functionality and efficiency metrics. The maximum estimation error for the hardware implementation was around 14.7%, with an average response time of 58.68 ms, a power consumption of 0.871 W, and a speed of up to 65.89 MBps. The hardware design and results obtained here can be applied in different tactile sensing systems such as robotics, prosthetic hands, biosensing, human-computer interfaces, and healthcare.
Download

Paper Nr:	69
Title:	Verifying Positivity of Piecewise Quadratic Lyapunov Functions
Authors:	Sigurdur Hafstein and Eggert Hafsteinsson
Abstract:	Continuous, piecewise quadratic (CPQ) Lyapunov functions are frequently used to assert stability for switched, cone-wise linear systems. It is advantageous to construct such Lyapunov functions in two steps: first a function is parameterized that is decreasing along all system trajectories, then it is verified whether this function is positive definite. Usually these steps have been performed using linear matrix inequalities (LMIs), but recently a linear programming (LP) approach for the first step has been suggested. In this paper we present a new algorithm to verify the positivity of CPQ Lyapunov function candidates, parameterized either with LMIs or LP. Further, we prove that the algorithm is non-conservative and will always be able to either assert positive definiteness of a CPQ Lyapunov function candidate or find a point where it is negative.
Download

Paper Nr:	116
Title:	Robust LiDAR-Based Parking Slot Detection and Pose Estimation for Shell Eco-Marathon Vehicles
Authors:	Miklós Unger and Ernő Horváth
Abstract:	This paper introduces the winning algorithm of the 2024 Shell Eco-marathon Autonomous Urban Challenge for autonomous parking. The task requires the vehicle to identify an available parking spot from multiple alternatives and precisely navigate into it, fully remaining within the designated area without touching any lane markings. Successful task execution requires not only reliable long-range detection of the parking space but also an accurate final orientation relative to the parking spot. To solve this task, we propose a novel method which relies on the combination neural networks and traditional point cloud processing methods. Since this is a highly specific problem tailored to the Shell Eco-marathon setting, and no publicly available solutions from other teams have been observed, our earlier algorithm serves as the primary baseline for comparison.
Download

Paper Nr:	141
Title:	From Algebraic Synthesis and GRAFCET to Logical Controller Design in ST Code (IEC 61131-3)
Authors:	Mathieu Roisin, Dimitri Renard, David Annebicque, Bernard Riera and Pierre-Alain Yvars
Abstract:	This paper addresses the problem of logic controller synthesis and the automatic generation of code compliant with the IEC 61131-3 standard, specifically Structured Text (ST) code. From a methodological perspective, two complementary approaches can be used to tackle this problem. The extensional approach explicitly represents the solution using models such as GRAFCET or Petri nets. In contrast, the intensional approach defines the solution space through a set of rules or constraints, without enumerating all possible solutions. Among intensional techniques, algebraic synthesis stands out as a formal method to derive controllers from specifications. We argue that combining extensional and intensional approaches leads to more efficient and robust controller design. To this end, we propose a hybrid workflow that integrates an extensional model (GRAFCET) with an intensional method (algebraic synthesis), enabling the automatic generation of IEC 61131-3 ST code. To support this workflow, we have developed two software tools: GReSTIC, for code generation and simulation, and BooG, for the algebraic synthesis and fusion of the two approaches. The proposed methodology is validated through a case study, demonstrating the automatic generation of reliable and standard-compliant ST code.
Download

Paper Nr:	144
Title:	Adaptive Output Control with a Guarantee of the Specified Control Quality
Authors:	Nikita Kolesnik
Abstract:	The paper presents a modification of the classical algorithm of adaptive output control in order to guarantee that the signal is found in the set specified by the developer at any moment of time. The paper extends the algorithm to systems with arbitrary relative degree. The aim of current research is to design a control law that will ensure that the error between the output and the reference signal will be in the following set. The effectiveness of the proposed method is illustrated with mathematical modelling.
Download

Paper Nr:	159
Title:	An Adaptive-Robust Strategy Design for Process Control
Authors:	Dumitru Popescu, Catalin Dimon and Pierre Borne
Abstract:	The paper presents a new design strategy for industrial process control applications. The adaptive-robust control approach considers both adaptive control advantages and robust control benefits; the connection between the two concepts preserves the imposed performances for the closed loop nominal control system. The combined adaptive-robust solution introduces the same integral criterion for parameters identification of the process and for the control algorithm design. An optimal integral criterion and an appropriate robust measure for degradation of the system performances due to variation of the model are introduced in an iterative mechanism. The theoretical approach presented in this paper is validated on a close loop control system, the application being developed in simulation. The proposed strategy is aiming to implement adaptive-robust control in practical process applications.
Download

Paper Nr:	165
Title:	Estimation of Rate-Dependent Hammerstein Model of Piezo Bender Actuator
Authors:	Lenka Kuklišová Pavelková
Abstract:	The paper presents a Hammerstein model of a commercial piezoelectric bender PL140 from Physik Instru-mente Co. The model consists of a nonlinear static part that describes the inherent hysteresis and a linear dynamic part that is represented by the auto-regressive model with exogenous input. The linear model parameters are estimated one-time using a particle swarm optimization algorithm. The rate-dependent nonlinear part is identified using input voltage data, along with a hidden variable that is obtained with the help of the inverted linear part. The experimental data are generated by a PL140 Simscape model with parameters set in accordance with catalog data.
Download