Development and Flight Test Evaluations of an Autonomous Obstacle Avoidance System for a Rotary-Wing UAV

Abstract

This paper presents the development and flight-testing of an obstacle avoidance system that can provide a rotary-wing unmanned aerial vehicle (UAV) the autonomous obstacle field navigation capability in uncertain environment. The system is composed of a sensor, an obstacle map generation algorithm from sensor measurements, an online path planning algorithm, and an adaptive vehicle controller. The novel approach of path planning presented in the paper is the integration of a newly developed receding horizon (RH) trajectory optimization scheme with a global path searching algorithm. The developed RH trajectory optimization scheme solves the local nonlinear trajectory optimization problem using approximated vehicle dynamics, maneuverability constraints, and terrain constraints within the finite range of the sensor. The global path searching by dynamic programming algorithm finds the shortest path to the destination to provide the initial guess to the RH trajectory optimization. The spline-based direct solver, Nonlinear Trajectory Generation (NTG), solves the RH trajectory optimization in real time and updates the solution continuously. The developed system is implemented within the Georgia Tech UAV Simulation Tool (GUST) and on the onboard computer of the Georgia Tech UAV test bed. Simulations and flight tests carried out for the benchmark scenarios and with sensor-in-the-loop flight tests demonstrated the viability of the developed system for autonomous obstacle field navigation capability of a UAV.