Uncertainty Velocity Obstacle Avoidance for sUAS Trajectory Planning in a 2D Plane

Abstract

To guarantee safe autonomous navigation of multiple small unmanned aircraft systems (sUAS) operating within a limited airspace, collision avoidance algorithms must be robust to uncertainties in aircraft states. The velocity obstacle (VO) concept is a popular avoidance algorithm which uses a collision cone to effectively determine if two objects will collide in the near future. The VO method is a reactive algorithm that allows for avoidance of dynamic obstacles. This paper proposes a novel approach, the uncertainty velocity obstacle (UVO) method, to enhance the decentralized VO collision avoidance method by addressing uncertainties in the position and velocity of moving obstacles. A scenario is presented to illustrate the utility of this method for an sUAS encountering other cooperating vehicles. In this scenario the vehicles use global positioning system (GPS) messages to communicate position and velocity between cooperative vehicles. Each vehicle uses these states in a constant-jerk Kalman filter to estimate other vehicles' positions and velocities. Numerical simulations show that UVO enhances a vehicle's ability to avoid collisions when operating in uncertain environments.