Hierarchical Adaptive Fuzzy Control of Mobile Robot in Dynamic Environment

This paper presents an obstacle avoidance method for mobile robot navigation. A fast and effective algorithm is presented for visual tracking and location prediction under dynamic environment. The picture image is first captured and the motion parameters of each object are extracted from the image. A sequence of images are then generated by using the adaptive fuzzy logic system and used to predict the next position and velocity of the moving object. The structure of the adaptive fuzzy logic system is similar to that of the Kaiman filter. The developed one-step ahead motion predictor can be used to predict the profile of the moving target. Monitoring the selected object using visual images, the vision system tracks the objects and adjusts the velocity and direction of the robot to avoid the objects. Using the predicted positions of the objects, an obstacle avoidance technique based on the adaptive fuzzy system is introduced. A hierarchical system structure is suggested to provide a systematic procedure to achieve target-directed navigation and it is implemented step by step integrating each part of the system designed separately. The effects of the individual subsystems are combined to increase the performance of the whole system. Computer simulations are presented for soccer robots in order to demonstrate the feasibility of the proposed algorithm.