Simulation Evaluations of an Adaptive Optimal Multiple Obstacle Avoidance Algorithm for Autonomous UAVs

Automatic obstacle avoidance is an essential requirement for UAVs to operate in obstacle-rich environment like urban areas that have led various researches for past decades. Among several aspects of the obstacle avoidance problem, finding the method of aggressive avoidance while maintaining the vehicle's motion within its safe flight envelope has been a typical interest for rotary-wing UAVs. As an approach to fulfill this interest, this paper presents an optimization-based method that generates optimal avoidance trajectory commands constrained by vehicle's maneuverability and other envelops limits in near real-time. Nonlinear trajectory generation (NTG) is used for the real-time trajectory optimizer. The proposed approach is realized in the nonlinear simulation tool, Georgia Tech Unmanned Aerial Vehicle Simulation Tool (GUST), and implemented into the onboard software of the GTMax test bed UAV. Part of the approach was tested successfully in a flight test performed by GTMax. Further extension to 3D obstacle avoidance is also discussed in this paper.