We investigated moving obstacle avoidance for Unmanned Aerial Vehicles (UAVs). We propose a two-stage approach for obstacle avoidance. Stage 1 involves a global path planner and Stage 2 uses a local potential field planner. By combining them in a hierachical manner, we are going to develop a complete system for avoiding stationary as well as moving obstacles for UAVs. The global planner uses Dubins curves (Ref. 1) coupled with a Kalman filter to predict the collision point and plan paths that are obstacle free. We demonstrate the efficacy of the system using simulated UAVs and real aircraft broadcasting their position using the ADS-B (Automatic Dependent Surveillance-Broadcast (ADS-B) (Ref. 2)) system. A local planner is also developed for avoiding obstacles that do not use ADS-B. This local planner uses vision for obstacle detection and potential fields for obstacle avoidance. The effectiveness of the global and local path planners are demonstrated in simulations.
展开▼
