Abstracts Track 2026

Area 1 - Robotics and Automation
Area 2 - Signal Processing, Sensors, Systems Modelling and Control

Area 1 - Robotics and Automation

Nr: 140
Title:

Optimization-Based Passivity Filtering for Variable Admittance Control
Authors:

Mustafa Akbaba, Andrea Cherubini and Arda Yiğit
Abstract: Variable admittance control enables robots to adapt their dynamic behavior across different tasks by varying stiffness and damping online. However, increasing stiffness may inject energy and violate passivity. This extended abstract presents two optimization-based filters that compute passive admittance profiles by minimizing their deviation from desired profiles under Lyapunov-based constraints. Simulations show improved admittance tracking and transient response compared with a baseline filter.
Nr: 166
Title:

Variable-Stiffness Mechanism Using a Telescopic Frame for a Cylindrical Elastic Crawler
Authors:

Yume Akino and Junya Nagase
Abstract: 1.Purpose: As public infrastructure continues to age, there is an increasing demand for robots capable of inspecting the inside of pipelines. Our laboratory has developed an amoeba-inspired cylindrical elastic crawler. Although this crawler can travel through small-diameter pipes, it can generate sufficient traction only over a 20-mm pipe-diameter range, from φ60 to φ80 mm. Therefore, this study aims to develop a variable-stiffness mechanism using a telescopic frame and thereby expand the traversable pipe-diameter range to four times that of the conventional crawler. 2.Design: A telescopic frame driven by a worm-and-rack mechanism was designed. Frame expansion and contraction change the effective length of the curved crawler belt, thereby varying its effective radial stiffness. The proposed mechanism was then applied to the cylindrical elastic crawler. 3.Findings: Prototype testing confirmed that the frame could expand and contract by approximately 70 mm, which satisfied the design requirement for accommodating a pipe-diameter range four times as wide as that of the conventional crawler. The crawler equipped with the proposed mechanism also successfully traveled through the target mock-up pipes. 4.Research Limitations: Although locomotion was demonstrated in mock-up pipes, performance in actual water pipes remains untested. Future work will evaluate the traversable pipe-diameter range using actual pipes. 5.Social Impact: The proposed mechanism may enable a single crawler robot to adapt to a wide range of pipe diameters, thereby contributing to more efficient inspection and maintenance of aging water-supply infrastructure. 6.Originality: This study introduces a variable-stiffness mechanism that changes the effective radial stiffness of a curved crawler belt by varying its effective length through frame expansion and contraction. The proposed mechanism offers a novel approach for adapting in-pipe inspection robots to various pipe diameters.
Nr: 167
Title:

Design of a Musculoskeletal Robot Finger toward Piano Tone Control through Muscular Relaxation
Authors:

Mai Yamada and Junya Nagase
Abstract: 1. Purpose This study develops a musculoskeletal robot finger that reproduces the anatomy and relaxation of the human index finger and investigates how joint relaxation affects piano keystroke dynamics. 2. Design The finger has DIP, PIP, and MP joints and six tendon-driven flexor and extensor paths. Its skeletal parts reproduce the structure and dimensions of human finger bones. Slender artificial muscles were arranged antagonistically to enable controllable relaxation. The extensor digitorum was placed dorsally, the flexor digitorum profundus and superficialis palmarly, and the interossei on both sides of the proximal phalanx. 3. Findings Prototype experiments confirmed that coordinated activation of the flexor, extensor, and dorsal interosseous muscles enabled piano key striking. Relaxed and non-relaxed conditions produced clear differences in keystroke dynamics, showing that joint relaxation changes the mechanical interaction between the finger and key. 4. Implications The results identify controllable joint relaxation as an important factor in piano keystroke dynamics. Although its causal relationship with tone remains unclear, the proposed finger provides a platform for investigating this relationship. Future work will analyze keystroke behavior and acoustic signals to quantify the effects of relaxation on tone. 5. Technical Impact The mechanism may contribute to performance robots and to assistive, rehabilitation, and prosthetic hands requiring compliant, human-like contact. 6. Final Comments A musculoskeletal robot finger with human-like skeletal geometry, tendon routing, and controllable relaxation was developed. Experiments showed that relaxation alters keystroke dynamics, providing a basis for clarifying the relationship among relaxation, keystroke behavior, and piano tone.
Nr: 170
Title:

Development of a Steering Mechanism Using an Air Cylinder for a Cylindrical Elastic Crawler
Authors:

Takashi Yamamoto and Junya Nagase
Abstract: 1 Pipelines such as water pipes require regular inspection to prevent accidents caused by damage. However, direct inspection of their interiors is difficult. Our laboratory has therefore developed a cylindrical elastic crawler inspired by amoeboid locomotion. In previous studies, the robot could travel in a desired direction through a T-shaped pipe, but its steering mechanism was structurally complex. This study proposes a simpler mechanism using an air cylinder. 2 The robot steers through T-shaped pipes by changing the number of driven belts. The air cylinder moves a slider between the belts and the robot frame, thereby controlling the engagement between the worm gear and the belts arranged at equal intervals around the circumference of the robot. 3 An operational test was conducted to determine whether the air cylinder could move the slider and change the belt engagement. The piston moved the slider to the intended position, and the engagement between the belts and the worm gear was successfully switched. 4 This study was limited to operational testing of the mechanism, and driving tests in actual pipes remain future work. Nevertheless, the results demonstrate the feasibility of controlling belt engagement with an air cylinder and indicate the potential for steering with a simpler structure. 5 This study contributes to simplifying steering mechanisms for self-propelled in-pipe robots. The proposed approach is expected to enable directional control in complex pipeline systems and improve the efficiency of pipe inspections. 6 In previous studies, a motor and gear mechanism was used to move the slider, resulting in a complex structure. The novelty of this study lies in replacing that mechanism with an air cylinder to control belt engagement. The results provide a basis for developing practical self-propelled in-pipe robots.
Nr: 171
Title:

Development of an Intelligent Partition Pole: Integration of Global Navigation, Local alignment, and Automatic Belt Connection
Authors:

Kazuma Kurosaki, Sho Yokota, Akihiro Matsumoto, Daisuke Chugo and Hiroshi Hashimoto
Abstract: Belt partition poles are commonly used to guide pedestrians and organize lines in airports, commercial facilities, and event venues. However, changing a line layout requires workers to carry, position, and connect multiple poles, creating physical burden and requiring direct manual work. To automate this process, we have developed an Intelligent Partition Pole (IPP) equipped with the omnidirectional mobility unit, the automatic belt connection mechanism, a ToF sensor for detecting the target pole, NFC readers and tags for identifying the connection slot, and an AR marker for global navigation. Previous studies evaluated the connection mechanism and local alignment separately, assuming that the poles were already located within the local-search range. This study integrates global navigation, local alignment, automatic belt connection, and belt deployment into a single system. The system consists of an IPPext, which extends the belt, and an IPPcatch, which receives and fixes it. An overhead camera detects the AR markers to estimate the position and orientation of each IPP. A central computer first guides the IPPext toward the vicinity of the IPPcatch and then switches to local alignment control. The IPPext aligns itself using the ToF sensor and moves around the IPPcatch until the NFC reader detects the tag indicating the connection position. It then stops, performs automatic belt connection, and resumes global navigation to reach the final position while deploying the belt. The integrated system was demonstrated in a 3.4 m × 3.4 m area using an overhead camera 2.7 m above the floor. In one proof-of-concept run, global approach, local alignment, automatic belt connection, and belt deployment were completed without human intervention. This result indicates the system's feasibility and supports future evaluation toward reducing direct worker involvement in partition pole installation.

Area 2 - Signal Processing, Sensors, Systems Modelling and Control

Nr: 165
Title:

Constructing Piecewise Quadratic Lyapunov Functions with Linear Programming for Continuous-Time Switched and Conewise Linear Systems
Authors:

Sigurdur Hafstein
Abstract: We present a method for constructing continuous piecewise quadratic (CPQ) Lyapunov functions for continuous-time switched and conewise linear systems using linear programming (LP). Key in our approach is the formulation of effective sufficient conditions for the copositivity of matrices via diagonal dominance. This formulation consists of linear constraints and can be expressed as an LP. One can show that the sufficient conditions are also necessary conditions for the existence of a Lyapunov function, given a sufficiently refined CPQ function. We provide an in-depth comparison between our new method and other computational methods in the literature, and provide extensive numerical experiments on various switched and conewise linear systems. In particular, we show that the proposed method is the most accurate of the LP-based methods for constructing Lyapunov functions and is a numerically competitive alternative to LMI-based methods.
Nr: 169
Title:

Deep Learning Model with Multi-Offset Gathering Ground-Penetrating Radar (GPR) Data for Water Leakage Detection in Urban Underground Networks
Authors:

Wenyao Lan, Théo Dezert, Benoit Vozel, Xavier Dérobert, Alexis Cothenet, Josias Lefevre and Amine Ihamouten
Abstract: The ground-penetrating radar (GPR) non-destructive technique uses electromagnetic wave propagation to detect dielectric and geometric characteristics related to water leakage zones without requiring specific water flow conditions. GPR antenna array configurations mainly include common offset gathering (COG) and multi-offset gathering (MOG), with COG being the conventional and commonly used method. As part of the Sci-Ty ANDANTE and FEREC JUMEAU projects, this study investigates the feasibility of analyzing MOG GPR data with two processing methods: Short-Time Fourier Transform (STFT) and f-k Transform (2DFFT), for water leakage detection. Simulated GPR data are processed into time/distance-domain (B-scan) images and spectral-domain representations, generating different grayscale image datasets. To evaluate water leakage detection performance under different antenna array configurations and signal processing methods, the datasets are separately classified using image classification model ResNet50. The three models using MOG data as input all achieved high performance (F1 scores and accuracy above 0.96), outperforming those using COG data (ranging from 0.67 to 0.92). Raw and filtered datasets are generated and applied as input. The performance of COG 2DFFT increases from 0.67 with raw data to 0.86 with filtered data, showing a strong effect of antenna propagation on COG data. This effect could not be highlighted for MOG configuration. Although using cleaned simulated data limits validation against complex real-world data, it provides an appropriate framework for evaluating the proposed processing methods. Future work will validate these methods with experimental GPR data already collected from two controlled test sites: Sense-City and JUMEAU test sites. In addition, the ANDANTE project includes real-case evaluations on streets, aiming to complement existing acoustic-based leakage detection methods used by water utilities and enhance detection accuracy.