Safe Path Planning for Unmanned Aerial Vehicle under Location Uncertainty

摘要

For the autonomous operations of unmanned aerial vehicles (UAVs), path planning satisfying collision avoidance plays an essential role in navigating through dynamic and uncertain environments. To this end, a probabilistic collision-free planning method based on Rapidly-exploring Random Tree (RRT) is presented in this paper. This method employs chance constraints to find a trajectory satisfying desired safety requirement for the UAV under location uncertainties for both the UAV model itself or obstacles. By assuming Gaussian distributions, the chance constraints arising from those uncertainties are formulated through converting dynamic, probabilistic constraints into equivalent static, deterministic constraints, namely the probabilistic bound. Also, a method named relative uncertainty is introduced to reduce different forms of uncertainties into a common one. On this basis, a chance constrained RRT planning algorithm is developed through combining formulated chance constraints with RRT. The feasibility of our proposed path planning method without violating the prescribed risk bound is well validated through numerical trials.