Search-based Path Planning and Receding Horizon Based Trajectory Generation for Quadrotor Motion Planning

This paper proposes a receding horizon based motion planning method, which allows a sensory-constrained quadrotor to dynamically plan obstacle-avoiding trajectories in unknown complex environments. First, a two-process search method is proposed to generate an initial feasible path satisfying the dynamics of the quadrotor. Second, the path smoothness is improved by solving a nonlinear optimization problem considering path safety and smoothness. Then, a uniform B-spline is used to interpolate the path with a receding horizon to achieve a safe and dynamically feasible trajectory with minimum trajectory time by solving an optimization problem. Finally, a time adjustment method is proposed based on the relationship between the distance of the B-spline trajectory and the obstacles. Extensive simulation results illustrate that the designed method doubles the safety range, defined as the minimum distance between the quadrotor and the obstacles, and consumes less than 70% of computational running time compared with the state-of-the-art. Outdoor flight experiments performed with a vision-based quadrotor show the satisfying performance of the motion planning approach.