Quadrotor Trajectory Generation in Dynamic Complex Environments

Recent advances in trajectory replanning have enabled quadrotors to navigate autonomously in static complex environments. However, navigation in dynamic environments still remains a significant challenge. In this paper, we present a trajectory planner that generates collision-free trajectories in environments with static and dynamic obstacles. A velocity obstacle (VO) based gradient field, called gradient velocity obstacle (GVO), is proposed to solve dynamic obstacles. The main improvement is that GVO maintains the original feasible set while ensuring computational efficiency. Using the output of GVO as an initial guess, a trajectory parameterized by a uniform b-spline is derived to avoid static obstacles. Multiple sets of comparative experiments show the validness and effectiveness of the proposed method.