Model Predictive Sample-based Motion Planning for Unmanned Aircraft Systems

This paper presents an innovative kinodynamic motion planning algorithm for Unmanned Aircraft Systems, called MP-RRT#. MP-RRT# leverages the idea of RRT# and the Model Predictive Control strategy to solve a motion planning problem under differential constraints. Similar to RRT#, the algorithm explores the map by constructing an asymptotically optimal graph. Each time the graph is extended with a new vertex, a forward simulation is performed with a Model Predictive Control to evaluate the motion between two adjacent vertices and compute the trajectory in the state space and the control space. As result, the MP-RRT# algorithm generates a feasible trajectory for the UAS satisfying dynamic constraints. Preliminary simulation results corroborate the proposed approach, in which the computed trajectory is executed by a simulated drone controlled with the PX4 autopilot.