Adaptive neural control of humanoid robot with unknown nonlinearities and uncertainties

This paper proposes an adaptive neural control scheme to deal with unknown nonlinearities and uncertainties in the humanoid robot manipulation. The deadzone nonlinearity in the detection channel is compensated in the control scheme. Moreover, adaptive laws are constructed to solve the control difficulty from unknown manipulated objects´ physical parameters. It is derived that all signals in the robotic system are kept bounded. Moreover, the motion error converges to the origin´s small neighborhood; the internal force error can be made arbitrarily small. Finally, experiment and simulation are carried out to illustrate the effectiveness.