Vision-Based Obstacle Sense and Avoid for Unmanned Aerial Vehicles
Project Award Date: 11-01-2017
The project aims to develop a new UAV sesne and avoid framework by solving the major challenges in the field. The project will be divided into three tasks, covering the problems of object detection and tracking, feature matching, and 3D depth recovery. The system adopts a pair of forward looking sterio cameras. We first detect the obstacles and intruders which appear in the field of view, then track these objects across the sequence and match them between the two views. Finally, we recover the depth of each object to that UAV and feed the distribution of these obstacles to the flight controller so as to make necessary avoidance maneuvers.
Primary Sponsor(s): Wichita State University.