Robust neural force control with robot dynamic uncertainties under totally unknown environment

In this paper, a new robust robot force tracking impedance control scheme that uses neural network as a compensator is proposed. The proposed neural com-pensator has the capability of making the robot to track a speciied desired force as well as of compensating for uncertainties in environment location and stiiness, and the uncertainties in robot dynamics. The neural compensator is trained separately for free space motion and contact space motion control using two different training signals. The proposed training signal for force control can be used regardless of the environment proole in order to achieve desired force tracking. Simulation studies with three link rotary robot manip-ulator are carried out to demonstrate the robustness of the proposed scheme under uncertainties in robot dynamics , environment position and environment stii-ness. The results show that excellent force tracking is achieved by the neural network. Recent developments in robot control have provided better productivity and eeciency in manufacturing industry. The necessity of achieving these beneets has drawn special attention in developing sophisticated control algorithms for robot manipulators. The constrained motion control is one of the areas that have been extensively studied. One of the main approach for constrained motion control is the well known impedance force control technique 1]. More recent focuses on impedance control research have been two folded : One is to have the force tracking capability that can follow the speciied desired force without any knowledge of environment location and stiiness. The other is to compensate for uncertainties in robot dynamics. Within this framework, many control algorithms have been proposed to tackle these problems 2]. Lasky and Hsia 3] have proposed the inner/outer loop control scheme that uses the integral control of the force error between desired force and actual force to modify the reference trajectory. Seraji and Col-baugh 4] have proposed the adaptive techniques of using force tracking errors to deal with uncertainties of environment stiiness and location. The authors in 5] have proposed the simple trajectory modiica-tion scheme with robust position control algorithm 6, 7] that compensates for uncertainties in robot dynamics and environment stiiness. In the paper 8], the authors also have developed neural network control schemes that can compensate for uncertainties in robot dynamics and environment stiiness based on impedance force control. In the latest paper 9], the authors have proposed a nonlinear impedance function that has capability of dealing with all the uncertainties when the environment …