Multi-Agent Planning and Game-Theoretic Decision-Making
This research focuses on strategic decision-making in multi-agent environments where autonomous vehicles interact under competitive or cooperative conditions. Leveraging concepts from game theory and differential games, the work aims to develop robust planning and control strategies that account for the intentions and possible deviations of other agents in the system.
A key application area is autonomous racing, where vehicles operate in close proximity and must anticipate each other’s actions in real time. By modeling the interaction between agents through Nash equilibrium, the framework identifies optimal policies that balance self-interest with environmental constraints.
The research also explores how agents can adapt when others deviate from Nash behavior, ensuring resilience and safety even under uncertainty. These methods enable interaction-aware motion planning, providing a principled foundation for dynamic, real-time decision-making in autonomous multi-vehicle systems.