Ant Colony Optimization Based on Local Optima Breaking Mechanism for Unmanned Vehicle Path Planning in Cross-Country Environment

Abstract

Intelligent vehicle is a preferable way for dealing with transportation problems. In cross-country environments, no apparent roads could be found, instead boscage, bumps and hollows, scattered materials make it hard to choose suitable paths. Path planning is a combinational optimization issue to find out a short path from many candidates. A combinational optimization algorithm was adopted to solve the problem. A multi-scale grid environment modeling mechanism was designed to model the unordered and irregular environment by a standard grid map. Then an ant colony optimization based algorithm was presented for finding out the most suitable path from the map. The algorithm introduced a local optima breaking mechanism based on early convergence judging process to avoid falling into local optima. Three criteria were designed for the process, say criteria of single grid, of static colony and of dynamic convergence. Feasibility and effectiveness of the method were verified by experiments.