Path Planning for Cooperative Sensing Using Unmanned Vehicles

Abstract

This work explores online path-planning for unmanned vehicles performing cooperative sensing. Much existing work employs receding-horizon optimization, where an objective function is repeatedly optimized over some short lookahead length. The use of receding horizon optimization often results in ad-hoc methods for dealing with the problem of myopic lookahead, where no value is visible in an agent’s planning horizon. This work examines the use of an algorithm for receding-horizon optimization that explicitly accounts for myopia by allowing for a variable lookahead length. Cooperation is maintained by ensuring that all agents plan to the same horizon, potentially with different strategies. This algorithm is used to develop trajectories for a team of unmanned vehicles searching for a target using a probabilistic framework. Simulation results are presented and discussed.