Abstract
This paper proposes a rapid trajectory planning algorithm using random search and a dynamics filter based on dynamic inversion. It can rapidly generate a collision-free trajectory for an unmanned aircraft in a highly constrained environment, including complex terrains with obstacles or waypoints. Conventional studies on trajectory generation mainly focus on global optimization; hence, the calculation cost grows exponentially with increased complexity. The proposed trajectory planner consists of two steps to effectively generate a collision-free flight trajectory. First, a sampling-based random search algorithm rapidly determines a collision-free pathway. However, the pathway generated by the random search is calculated geometrically, so the variations in the actual flight are difficult to follow. Therefore, the second step, named dynamics filtering, modifies the initial pathway to a dynamically flyable trajectory with a one-time-only integral calculation. The dynamic inversion technique is utilized to ensure flight-worthiness. In this paper, the implementation and computer simulation results of the proposed trajectory are described. The effectiveness of the proposed method is verified by a six-degrees-of-freedom flight simulation for a Mars exploration aircraft model developed by JAXA and associated research institutes.
