Reactive Vision Based Obstacle Avoidance with Camera Field of View Constraints

Abstract

The lack of reliable, low power, low weight sensors make automatic obstacle avoidance for fixed wing micro air vehicles (MAV) a difficult problem. Traditional sensors for obstacle detection such as radar are too heavy and too large for MAVs and generating paths around detected obstacles is non-trivial. EO/IRCameras are small and lightweight providing a viable alternative to large sensors. We assume that vision processing provides object segmentation and a range estimate to nearby obstacles. We propose a nonlinear guidance law that directly reacts to image data by pushing the obstacles to the edge of the camera field-of-view, and thus avoiding collision. As the MAV passes the obstacle, the original trajectory is resumed. This method can be used in conjunction with high level path planners.