Population of a range bearing map for local obstacle avoidance using monocular vision

This paper presents a method for navigation of a small unmanned aerial vehicle (UAV) through an unsurveyed static environment using a single forward pointing camera and a GPS corrected inertial measurement unit. The size and maneuverability of small UAVs allows for low altitude flights in complex environments. Without accurate estimates of obstacle locations, successful navigation is not possible. Using camera measurements of bearing and optical flow and estimates of vehicle motion, a range bearing map is created. The range bearing map stores estimates of the closest obstacle in a particular region, the information necessary to perform local obstacle avoidance. By discretizating the area around the vehicle by angle and forcing each region to contain an estimated range, complex environments can be modeled. Results of two-dimensional simulations are presented using a potential field method for obstacle avoidance and navigation.