Vision based object follower automated guided vehicle using compressive tracking and stereo-vision

Abstract

Integration of a visual sensing system plays a vital role in automated navigation by providing a sensing ability of the surrounding environment. The problem of object following is challenging due to changes in appearance that can occur due to motion, pose, illumination and occlusion. The real-time implementation of a computer vision based object following system is presented in this paper. The position of the object to be followed is determined by processing a real time image feed from a calibrated stereo-camera. The method incorporates compressive tracking and stereo-vision based disparity mapping boosted with relocation of the tracking window using Speeded Up Robust Features (SURF). The proposed algorithm runs in real-time and performs favorably in terms of computational efficiency, accuracy and robustness.