Real World Collision Avoidance for Humanoid Robots

The objective of this thesis is to develop a methodology for motion generation of humanoid robots that enables them to work in the real world. The basic concept is to integrate both methods: reactive control and planning and to use an architecture which utilizes both methods depending on the environment in order to compensate their drawbacks.The first target of the thesis is to realize reactive control which comprises target reaching and collision avoidance control for collision-free motions independent of planning. Since the reactive target reaching control has been developed mainly derived from methods originally targeted for industrial robots, the thesis focuses more on the reactive collision avoidance.