Vision-based reactive multiple obstacle avoidance for micro air vehicles

Traditional sensors for obstacle detection on aircraft, such as radar, are too heavy for fixed wing micro air vehicles and make obstacle detection and avoidance a difficult problem. EO/IR cameras are small and lightweight enough to offer an alternative to heavy sensors. Vision processing available on a ground station provides range and bearing to nearby obstacles. We propose a nonlinear guidance law based on movement of obstacles in the camera field-of-view to ldquopushrdquo the obstacles to the edge of the camera field-of-view and thus avoid collision. As the MAV passes obstacles, the original course is resumed. The guidance strategy is demonstrated in simulation and flight test. This reactive method intended for use in conjunction with high level path planners.