A Hybrid Intelligent Navigation System for Autonomous Vehicles in Contested Environments: Integrating Modified Ant Colony Optimization with Threat-Weighted A*

Abstract

The challenges associated with autonomous navigation in combat scenarios include communications degradation of the navigation sensors, dynamic hostile threats, and the necessity for real-time optimization of multiple objectives. Current mapping methods either do not provide enough speed for proper replanning or they do not deliver a globally optimal path under persistent uncertainty. In this paper, we present a novel two-tier, hybrid path planner that combines globally determined routes using an altered version of the ant colony optimization (ACO) algorithm in conjunction with locally created, reactive routes generated by a threat-weighted A-star algorithm (A*). The significance of this hybrid architecture is that through the separation of strategic and tactical planning processes, both path length, threat exposure, and the ability to travel over terrain can all be concurrently optimized. Monte Carlo simulation results from experimental validation conducted on a 30x30 grid with dynamic threat activation yields average path efficiencies of 46.0% with a threat exposure rating of 0.90 and computational time of 0.40s per planning cycle based on theoretical analyses of the hybrid method. Convergence issues were identified in the implementation of the hybrid method, and additional refinement is warranted. The Standard A* algorithm produced the highest path efficiency of 70.7% but had a computational time of only 0.01s. The Threat-aware A* algorithm produced the lowest threat exposure ratings of all three algorithms, with a maximum value of 0.38. Thus, it was found that both algorithms set the groundwork for future development of hybrid navigation systems in contested environments.