A load-adaptive controller for humanoid robots

A crucial problem for humanoid robots is to adapt to the uncertain external environment. Since load is one of the most variable parameters in a humanoid robot and affects the high performance in trajectory tracking, strict requirements for load adaptive controllers have been put forward. In this paper, we present a high performance load-adaptive controller based on BP neural network. The adaptive controller adapt robotic nonlinear systems and the coefficients in the controller can be tuned automatically on-line. The effectiveness of the adaptive controller is confirmed by experiments on BHR-5 humanoid robot.