Upper Body Tracking for Human Robot Collaboration

08.05.2009, Abschlussarbeiten, Bachelor- und Masterarbeiten

In the Cognitive Factory -- one of the central demonstrator of the excellence cluster of CoTeSys -- the project JAHIR (Joint Action for Humans and Industrial Robots) aims to integrate human workers in a shared workspace with industrial robots. Industrial robot and human worker should work together on a common construction task.

In such a scenario, the safety for the human worker plays an important role. The robot needs to know the exact position of arms, hands, the head, and the body to interact safely with the human. Beside the safety aspect, the movements of the human worker can be analysed in such a hybrid assembly scenario. The knowledge about the trajectories and the sequence of movements can e.g. be used to analyse the efficiency of the collaboration or to anticipate next steps of the human. Therefore, the aim in this work is to track the upper body of a human co-worker with multiple cameras in realtime. The thesis will be based on a sophisticated software framework for model-based visual tracking. This framework offers already a large variety of algorithms that can be used to solve the task. Within the thesis new algorithms should be integrated in the structure of this framework. To cope with the computational complexity, the computational power of graphics hardware (GPU) will be used. More details are available under: http://www6.in.tum.de/Main/StudentProjectsFullBodyTracking

Kontakt: Claus Lenz (lenz@in.tum.de)