SHORT-TERM CONFLICT RESOLUTION FOR UNMANNED AIRCRAFT TRAFFIC MANAGEMENT

Ensuring safety and providing timely conflict alerts to small unmanned aircraft is important to their integration into civil airspace. We propose several algorithms for short-term conflict avoidance as part of an automated traffic management system. The goal is to balance aircraft safety and efficiency subject to environmental and aircraft uncertainty. The proposed controllers generate advisories for each aircraft to follow, and are based on decomposing a large Markov decision process and fusing their solutions. We separate the problem into pairwise encounters that are solved offline. The solutions to these encounters are then combined online to produce a locally optimal solution using an iterative search technique. As a result, the methods scale well and the global problem can be solved efficiently. We demonstrate the controllers by evaluating them against baseline algorithms in simulation.