Fuzzy control for obstacle avoiding in mobile robots using stereo vision algorithms

Abstract

Stereo vision has been an active research field for several scientific communities. Most of the reported results have considered pure image treatment algorithms. Nevertheless, there are few methods that combine stereo vision with automatic control algorithms to regulate the position of mobile robots. In this paper, the combination of stereo vision image treatment algorithm with a type-1 fuzzy logic controller. This controller is designed to avoid the collision of the mobile robot with obstacles located within the space where a robot is moving. Indeed, the controller was designed to accomplish two tasks, the obstacle avoidance and the general movement in the region of activity for the mobile robot. The stereo vision algorithm was successfully implemented to calculate the distance between the mobile system and the obstacle (based on block-matching algorithms: exhaustive and pyramidal) as well as the attack angle between the same hurdle and the rotation performed by the robot (using the optical flow algorithm). The fuzzy controller was designed using simple rules based on the observed trajectories performed by the mobile robot within a model scenario used to test the controller performance. The complete system was implemented into a real mobile robot to prove the efficiency of the mixed scheme using stereo vision and fuzzy control subsystem. The resulted mobile system can avoid the obstacle in all experiments independently of the relative position between them.