Learning Real-Time A* Path Planner for Sensing Closely-Spaced Targets from an Aircraft

Abstract

This work develops an any-time path planner, based on the learning real-time A* (LRTA*) search, for generating flyable paths that allow an aircraft with a specified sensor footprint to sense a group of closely-spaced targets. The LRTA* algorithm searches a tree of flyable paths for the branch that accomplishes the desired objectives in the shortest distance. The tree of paths is created by assembling primitive turn and straight sections of a specified step size. The operating parameters for the LRTA* search directly influence the running time and path-length performance of the search. A modified LRTA* search is presented that terminates when there has been no improvement in the path for some number of iterations, resulting in a path planner that provides short-distance paths in short running times.