Efficient Trajectory Planning via Projection onto Convex Sets for Unmanned Aerial Vehicle

Safe UAV navigation is challenging due to the diversity and dynamic nature of obstacles. Although existing works have made progress in handling either static or dynamic obstacle scenarios, there remain limitations in simultaneously addressing both types of obstacles. To overcome the challenges, we propose a real-time UAV trajectory planning framework for safe navigation in complex environments with both static and dynamic obstacles. We firstly propose a corridor generation algorithm based on convex projection in order to generate a static trajectory. Secondly, for dynamic obstacles, we use the static trajectory as a reference trajectory and applies the Velocity Obstacle (VO) method to filter safe velocity intervals. We further incorporate Model Predictive Control (MPC) method to generate real-time dynamic trajectories, which effectively handles uncertainties in dynamic environments. Simulation results demonstrate that our framework outperforms existing planning frameworks in terms of real-time performance and safety.