Adaptive Fuzzy Optimal Control of Modular Robot Manipulators Systems via Integral Reinforcement Learning-Based Value Iteration Algorithm

Abstract

A novel adaptive fuzzy optimal compensation control approach is developed for modular robot manipulator (MRM). By incorporating integral reinforcement learning (IRL) to value iteration (VI), the adaptive fuzzy optimal control approach effectively addresses the trajectory tracking problem without accurate dynamic model information, and break the limitation of traditional policy iteration (PI). Moreover, this article approximate cost function with fuzzy logic system (FLS), which simplify the solution of the optimal control policy and better capture the important features of the cost function.