RRT-based path planning for fixed-wing UAVs with arrival time and approach direction constraints

This paper proposes a path planning algorithm based on rapidly-exploring random trees (RRTs) for fixed-wing unmanned aerial vehicles (UAVs). The algorithm uses a pre-defined motion primitive set to extend the tree, and can be designed to reflect the dynamic capability of a target aircraft. The proposed method produces collision-free, dynamically feasible, and smooth curves. Furthermore, the algorithm can determine the approach direction of each initial and goal position and the arrival time at the goal position by generating an inertial speed command. Estimated dynamic information, including roll, heading, required thrust, and aerodynamic force are also output. The algorithm is validated by performing several simulations and comparing its performance with the response of a nonlinear six-degrees-of-freedom simulation. The results show that this algorithm can be successfully applied to fixed-wing UAV applications.