Uniform ultimate boundedness of a neural PD controller for robot manipulators with uncertain parameters.

In this work a stable neuro PD controller for set point control of planar robot manipulators in task space is proposed. The gravity terms and the robot Jacobian matrix are assumed unknown. Gravitational terms are approximated using Radial Basis Functions Neural Network with task or joint information feeding the activation functions and with on-line real-time learning. Sufficient conditions for the bound on the estimated Jacobian matrix and stability condition for the feedback gains are presented to guarantee that all the closed loop signals are uniformly ultimately bounded. Experimental results in a two degrees of freedom robot are presented to evaluate the proposed controller.