Multi-Sensor Integration in Robotic Systems Using On-Line Trajectory GenerationAlgorithms

• Applicant: Professor Dr.-Ing. Torsten Kröger
• Subject Area: Automation, Mechatronics, Control Systems, Intelligent Technical Systems, Robotics
• Term: from 2009 to 2012
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 152926022

This proposal reports about results that have been achieved during the currently funded research project. The main objective of the ongoing research is to extend the theoretical framework for on-line trajectory generation algorithms. In addition to these conceptual and theoretical extensions, real-world experimental results are presented and discussed. The extended methodology could successfully be integrated in the control architecture of a seven-degree-of-freedom robot arm as well as in the control system of a humanoid robot, such that these systems are now capable to react instantaneously to unforeseen events and to switch between frames and control spaces at arbitrary instants of time. In order to furthermore exploit the high potential of this framework, a continuation is proposed, and three main objectives are suggested: 1. Extending the theory furthermore, such that workspace limits can be used to determine maximum allowed values for desired target states of motion. 2. The extension that has been achieved so far enables us to use the resulting algorithms as control submodules within hybrid switched systems. In order to obtain benefits for other fields and to utilize the proposed control scheme in other applications, a proof of stability is planned to be derived. 3. Assume, humans and robots share the same workspace, and that the workspace is surveyed by high-speed camera systems. If the framework of on-line trajectory generation is furthermore combined with a sensor-based on-line motion planning system, extremely low response times can be achieved, such that humans may not be able to touch the end-effector of the robot, because the robotic system can react to the motions of the human with a minimum latency (< 3 ms). Besides the new theoretical derivations that are proposed here, real-world experiments are planned for all three objectives in order to significantly contribute to the highly relevant field of safe human-machine interaction.

• DFG Programme: Research Fellowships
• International Connection: USA
• Host: Professor Oussama Khatib