SOAR - Stereo Obstacle Avoidance Rig

OKSI and Professor Frank Dellaert of Georgia Institute of Technology (Georgia Tech) are teaming up to develop an ultra-low cost passive low-SWaP spherical situation awareness sense/avoid system based upon monocular stereo vision (i.e., stereo-from-motion) for small UAS platforms operating within the NAS. When flying close to the ground (e.g., during takeoff and landing) obstacles such as cars, trees, buildings, power lines, people, and so on are not equipped with beacons. In this setting, the ability to actively detect obstacles within the environment in real-time and to take evasive maneuvers to avoid collisions is a required capability for safe operation in the NAS. Currently, there are no existing technologies that sufficiently address the sense/avoid problem associated with operation of small UAS platforms (<55lbs) operating within the NAS. To this end, OKSI is developing the Stereo Obstacle Avoidance Rig (SOAR) that will provide a complete solution to the sense/avoid problem for small UAS platforms. The SOAR system utilizes a video stream from a distributed aperture array of cell phone cameras combined with state-of-the-art single-camera stereo vision algorithms in order to construct accurate 3D environmental maps in real-time.