Randomized path planning with deceptive strategies

In this paper we study a non-cooperative zero-sum game where one player performs reconnaissance while the second player constantly observes the first. This game has implications for teams of UAVs operating within aural and visual detection range of threat forces. In particular, the threat can potentially react dynamically to UAV observations and endanger future movements. We propose a specific behavior essential to an optimal policy for a team of agents, and create a randomized algorithm inspired by these heuristics. Implementation of this path planner onto a team of autonomous helicopters demonstrated the utility of the algorithm in real time applications.