Hyper-Heuristics for Online UAV Path Planning Under Imperfect Information

Abstract

Hyper-heuristic techniques are problem independent meta-heuristics that automate the process of selecting a set of given low-level heuristics. Online path planning in an uncertain or unknown environment is one of the challenging problems for autonomous unmanned aerial vehicles (UAVs). This paper presents a hyper-heuristic approach to develop a 3-D online path planning for unmanned aerial vehicle (UAV) navigation under sensing uncertainty. The information regarding the state of a UAV is obtained from on-board sensors during the execution of a navigation plan. The trajectory of a UAV at each region is represented with B-spline curves, which is constructed by a set of dynamic control points. Experimental study performed on various terrains with different characteristics validates the usage of hyper-heuristics for online path planning. Our approach outperforms related work with respect to the quality of solutions and the number of feasible solutions produced.