Stereo vision based robots: Fast and robust obstacle detection method

摘要

In this paper we present a new obstacle detection method, based on stereo vision, without combination with any other kind of sensors. The proposed method uses a differential image transform algorithm to gain robustness against illumination changes. This method increases the speed of program execution while keeping the performance of stereo vision algorithm in term of accuracy in the same level with the previous algorithms. Moreover, we implement this method into a stereo vision based robot while adding some new features to widen the depth detection range. With the help of the proposed method, the robot detects obstacles between 25cm to 400cm from robot cameras. The result shows the robot has the ability to work in a wide variety of lighting conditions, while the stereo vision part of the robot does the depth detection computation with the speed of 30FPS.