Real-time safety-monitoring for mobile robot navigation on an industrial automation computing platform

Master thesis defense

Eslam Salah Zaki Elshiekh

University of Freiburg / Robert Bosch GmbH

Thursday, November 17, 2022, 10:00 - 11:15

Room 01-012, Georges-Köhler-Allee 102, Freiburg 79110, Germany

Safety-guaranteed fast navigation is one of the key components for mobile robots. This master thesis is based on an existing safety-monitoring approach which considers pedestrians not as static obstacles but instead models the dynamic behavior of pedestrians to enable faster navigation: It stops the robot only when the pedestrian reachable sets and the sets of the predicted robot positions intersect. However, this computational expensive algorithm requires a high-performance computer and even then does not run in real-time. This thesis not only optimizes this algorithm but also continuously monitors its execution time, so that it could be safely deployed on an embedded control platform (ctrlX AUTOMATION) from Bosch Rexroth. To demonstrate its modularity it has been integrated into the ros2_control framework, which is an open-source software package for real-time control applications in ROS 2 (Robot Operating System). Moreover, this thesis integrates a pedestrian detection module (DARKO) with the improved safety monitoring approach. The experiments on the real mobile robot (ActiveShuttle) demonstrate the efficacy of the approach: The safety-monitoring approach shall run at 10 Hz to qualify as real-time and process 25 pedestrians. While the existing approach requires 13 seconds on the ctrlX-AUTOMATION platform, the improved approach needs only 3 ms, and thereby, clearly is enabling a safety-guaranteed fast navigation on a low-cost embedded platform.


Also online via Zoom:
Meeting ID: 627 9173 7415
Passcode: syscop2021