PREDICTIVE FUZZY CONTROL OF AN AUTONOMOUS MOBILE ROBOT WITH FORECAST LEARNING-FUNCTION

This paper deals with the drive control of an autonomous mobile robot. An autonomous mobile robot is one of the intelligent robots that need abilities to recognize and to adapt to surrounding environment. We propose a new approach to meeting these needs. This approach is based on a forecast learning fuzzy control. The environment can be classified into several characteristic patterns and our robot has sets of control rules for each pattern beforehand. The robot integrates these sets into a single set using degrees of matching between the current environment and each pattern. The robot forecasts whether it will drive safely or not by prediction, by using the integrated control rules. The robot considers the results of the forecast, and then adjusts the conclusion parts of the integrated control rules in order to drive more safely in such an environment. In this paper, to find the efficacy of our new approach, the simulation results of the drive control of the robot and the experimental results on indoor routes are shown.