Multimodal Error Detection in Robotic Assembly (SA/FP/IDP/MA)

Summary

Beschreibung Motivation The iwb researches the robotic assembly of deformable linear objects such as cables. Errors appear in different signals – acoustics, force/torque, camera images, and poses. Every single modality detects only part of them, so a synchronized multimodal approach is needed. Objective Integrate acoustic, force/torque, vision, and pose-based error detection into one system that handles errors autonomously. Work Packages • Literature review on multimodal sensor fusion • Develop a Synchronization strategy for the asynchronous sensor streams • Define a common interface and event format for all detectors • Decide on a fusion logic for cross-modal error hypotheses • Link fused detections to recovery skills • Benchmarking single-modality vs. multimodal detection and validation at the robot cell Voraussetzungen Requirements • Structured way of working • Good programming skills (Python, ideally ROS) • Independent problem-solving skills Contact M. Sc. Celina Dettmering Department Assembly Technology and Robotics Celina.Dettmering<script>document.write('@');</script> <noscript>(at)</noscript>iwb.tum.de