Vision-based navigation frame mapping and planning for collision avoidance for miniature air vehicles

This paper presents a vision-based navigation frame mapping and path planning technique for collision avoidance for Miniature Air Vehicles. A depth map that represents the range and bearing to obstacles is obtained by computer vision. Based on the depth map, an extended Kalman Filter is used to estimate the range and bearing. Using this information, a map, constructed in polar coordinates, is created in the navigation frame of the MAV. The Rapidly-exploring Random Tree algorithm is employed to find a collision-free path in the navigation frame. The proposed algorithm was successfully implemented in both simulation and flight tests.