Prof. Karsten Berns
A Biological inspired Approach for Bipedal Locomotion
Bipedal locomotion is still an open research area. Till now the locomotion skills and the energy consumption of a human size biped is still not comparable with human walking and running. In literature there are two research directions which examine different solutions. The conventional bipedal control approaches focus on the modeling of the multi-body dynamics and derive walking strategies (e.g. ZNP, optimal control). The second class of control approaches try to transfer biological and biomechanical aspects of human locomotion for bipedal walking robots. Among them, the Bio-Inspired Behavior-Based Bipedal Locomotion Control (B4LC) is developed by the Robotics Research Lab (RRLab) at the University of Kaiserslautern. The concepts tries to transfer the principle of combining a fixed number of synergies and reflexes for different walking phases to achieve human like walking behavior. The B4LC approach is used to develop certain walking skills, like walking initiation, cyclic walking, walking termination, curve walking and upslope walking. To enhance the robustness of bipedal walking in more challenging environments, a learning scheme is further implementedin the B4LC by using the Swarm Particle Optimization and the reinforcement learning method. The advanced locomotion skills are extended with proposed learning scheme, such as walking with different velocities, compensation of external pushes, and stepping over large obstacles. Experiments are performed with the help of a dynamic simulation as well as with a test leg.
قزوین، خیابان دانشگاه، بلوار نخبگان، بلوار پژوهشگران، دانشکده کامپیوتر و فناوری اطلاعات، دبیرخانه کنفرانس، اتاق 205
کد پستی :