The comparison research of robot control using BP and CMAC neural network

Discusses two manipulator control strategies using neural nets, BP and CMAC. No prior knowledge of robot dynamics and environments is required when designing the NN controller, because the BP and CMAC have the ability of self-organization and self-learning. The controller consists of two parts,one is a conventional feedback loop, it forms the primary control torque, another is a NN compensator, it forms the torque modification terms, to guarantee the desired trajectory. A two joints manipulator is chosen as the controlled system. The comparison is made between the two NN controllers respect to the learning speed, the effect of noise and load, and a non-repetitive task. The simulation shows that two NN controllers have better performances than conventional controllers, and the CMAC controller has the more rapid convergence and simple configuration, so that it could be used in a real-time manipulator control system.