Adaptive Traffic Following for Decentralized Autonomous Aircraft

For my 2025 summer internship project at NASA Ames, I have developed a decentralized multi-agent aircraft simulation. Each agent has a Dijkstra-based path planner which encourages traffic following. The path planner then passes waypoints to a lower-level model predictive path integral controller (MPPI) with repulsion-field based collision avoidance which encourages the agent to move towards the waypoint safely. Here is a demo video of the simulation in action. More details coming soon.