Real-time target tracking and obstacle avoidance for mobile robots using two cameras

Abstract

This paper proposes an image processing approach for real-time target tracking and obstacle avoidance for mobile robot navigation in an indoor environment using two cameras. Several image processing techniques which include the averaging filter, edge-detection, erosion and dilation, and color segmentation are combined to find the target and obstacles. Then one can compute the angular position of the detected target and obstacle related to the mobile robot. The two cameras are utilized to calculate the relative distance of the target and obstacle from the mobile robot. According to the distance, one can determine the relationship of the target and obstacle to the mobile robot. The best target tracking path and obstacle avoidance path can be determined by different behavior modes. Therefore, the mobile robot plans a collision-free and successful track target to complete the patrol routine. Finally, the practical experiments demonstrate the feasibility and effectiveness of the proposed schemes.