Adaptive fuzzy controller for robot navigation

摘要

In this paper an adaptive fuzzy controller based on the Takagi-Sugeno method is presented and a learning procedure for it is derived. First, it is shown that the Takagi-Sugeno controller can be modeled as a generalized (extended) form of the "conventional" radial basis function (RBF) network. A supervised learning procedure for such a controller is then developed. The learning capabilities were tested on the nonlinear function approximation problem and the results showed that the learning speed was higher than that of conventional RBF network. This adaptive controller was applied to the tasks of robot obstacle avoidance and wall following and satisfactory performance was also achieved.