Fuzzy logic based approach for robotics systems control. stability analysis

This paper presents an adaptive fuzzy control approach for complex tasks involving robot-environment interaction. Implementation of this approach is based on the design and optimization of fuzzy logic controller (FLC) performed in two stages. In the first stage, the FLC parameters are trained and optimized offline using a rapid prototyping algorithm combined with a method based on Solis and Wetts algorithm. The latter algorithm allows convergence of the cost function to its global minimum by using a local search algorithm which is a randomized hill-climber with an adaptive step size. For convenience of analysis, the structure of the FLC is divided into multi-input-single-output (MISO) controllers. In the second stage, an online learning of the FLC is then implemented into the proposed control structure. Robustness of the proposed approach is shown by the stability analysis based on the Lyapunov method. To show the performances of the proposed approach, simulations are carried out on a 3DOF robot performing contour following under force constraints.