Adaptive Position/Force Controller Design Using Fuzzy Neural Network and Stiffness Estimation for Robot Manipulator

Abstract This paper proposes an adaptive hybrid position/force control approach using fuzzy neural networks (FNNs) for a robot manipulator with joint friction compensation. The dynamics model and system uncertainties are estimated by FNNs. For force tracking control, an adaptive impedance controller is employed with an online stiffness estimator, wherein the stiffness of the contacted environment is estimated using a gradient descent algorithm. The adaptive update laws of the FNNs and the stability of the controller are obtained using the Lyapunov stability theorem. Finally, the proposed adaptive hybrid controller is implemented on the AR605, a 6-axis articulated robot manipulator manufactured by the Industrial Technology Research Institute (ITRI). The effectiveness and robustness of the proposed control strategies are verified by the simulation and experimental results.