UAV 3D path and motion planning in unknown dynamic environments

Unmanned aerial vehicles (UAVs) are used in increasingly complex environments. This paper deals with path and motion planning, based on the local perception of the UAV, generating autonomously a safe 3D path in dynamic cluttered environments under different constraints. The latter are formulated in a multi-criteria real-time optimization process that aims to minimize an objective function while satisfying the dynamic constraints of the UAV and the safety requirements. The considered constraints impact the velocity and the acceleration of the UAV, but also its curvature along the trajectory. Safety constraints are ensured by a minimal-security distance and by locally shifting the heading of the UAV to avoid obstacles while maintaining attractiveness to the goal location. Simulation results under ROS/Gazebo show the effectiveness and robustness of the approach even in highly cluttered environments. We study the effect of several parameters on the optimization process and the resulting trajectories.