Designing Ping-Pong Player Robot Controller with ANFIS

Abstract

Designing of robot controllers is the most commonly used in the engineering fields. In this matter, intelligent designing of controller with minimum human intervention is a major challenge. We chose ping-pong player robot to run a controller training algorithm on it. Our Algorithm has 2 main sections. 1st we produce a pattern of the best racket path for each ball path. In the 2nd section we will use the produced pattern to train a Nero-Fuzzy inference system. In this article we study 2nd section. We have chosen Adaptive Neuro-Fuzzy Inference System as our controller due to its ability of training and work with complexity. We train the ANFIS controller with existing pattern. To train ANFIS we will use evolutionary algorithms to optimize ANFIS controller efficiency by adjusting membership functions shapes, and inference engine rules.