Probabilistic multi-hypothesis tracker for multiple platform path planning

Abstract

This study considers the problem of automatically coordinating multiple platforms to explore an unknown environment. The goal is a planning algorithm that provides a path for each platform in such a way that the collection of platforms cooperatively sense the environment in a globally efficient manner. The environment is described by a spatially non-homogeneous priority function. The method samples this function to produce a discrete collection of locales that the platforms use as waypoints. The key feature of the method is to treat the assignment of locales to platforms as a target tracking problem and to use the probabilistic multi-hypothesis tracker (PMHT) as a method of performing multi-platform batch data association. This paper introduces the PMHT path planner (PMHT-pp) and compares this algorithm as a method of performing multiple platform batch data association with the Genetic Algorithm to solve the modified multi-travelling salesman problem.