A Software Library for Autonomous Obstacle Field Navigation

Operating unmanned air vehicles (UAVs) in complex or cluttered environments requires advanced guidance and navigation software. This software must quickly reason about and react to the sensed environment without requiring too much computing power.

libOFNlibOFN is designed to fill the role of route planning in the presence of obstacles, taking into account the dynamic limits of the vehicle [1]. It has been validated in autonomous flight tests by the US Army Aviation Development Directorate - Aeroflightdynamics Directorate [2] and against near-optimal benchmarks [3]. Using 3-dimensional sensor readings and a destination waypoint as inputs, libOFN generates a safe trajectory and reacts to newly-detected obstacles while flying this plan. In addition to the route planning algorithm, libOFN contains other elements necessary for planning. These include a terrain representation of known and unknown space as well as a spline-based trajectory generator that accounts for aircraft dynamic limits. A reactive planning module ties these elements together and manages the response to newly sensed obstacles and changes in the environment.

The U.S. Army Research, Development and Engineering Command's Aviation and Missile Center successfully demonstrated low-level autonomous behaviors in a flight over the Diablo Range. This demonstration is critical to the next generation of technological advances in military rotorcraft: obstacle field navigation and safe landing area determination.

Terrain sensing, statistical processing, risk assessment, threat avoidance, trajectory generation, and autonomous flight control were performed in real-time during the two-hour test flight. Full news article can be found here