Design and Evaluation of a Multi-Agent Software Architecture for Risk-Minimized Path Planning in Human-Robot Workcells

Title data

Werner, Tobias ; Riedelbauch, Dominik ; Henrich, Dominik:
Design and Evaluation of a Multi-Agent Software Architecture for Risk-Minimized Path Planning in Human-Robot Workcells.
In: Schüppstuhl, Thorsten ; Franke, Jörg ; Tracht, Kirsten (ed.): Tagungsband des 2. Kongresses Montage Handhabung Industrieroboter. - Berlin : Springer Vieweg , 2017 . - pp. 103-112
ISBN 978-3-662-54441-9
DOI: https://doi.org/10.1007/978-3-662-54441-9_11

Project information

Abstract in another language

Close quarters human-robot collaboration promises the integration of human flexibility with robot precision, endurance, and strength. However, such collaboration requires a monitoring solution for the shared human-robot workspace to drive the robot manipulator. Recent research has already developed a variety of individual software components for monitoring solutions, from object detection over reconstruction to path planning. In our contribution, we implement existing components for multi-camera reconstruction, risk estimation and path planning in the form of concurrent agents and we connect these agents through a shared blackboard in order to realize risk-minimal path planning for a robot manipulator. We discuss our software architecture, and we evaluate the final software in an example application of human-robot collaboration. In conclusion, our contribution enables risk-minimal path planning on an industrial manipulator at a 10 Hz update rate.