Strategically Robust Aggregative Games

Researchers Andreas Feik, Nicolas Lanzetti, Saverio Bolognani, Florian Dörfler, and Dario Paccagnan from ETH Zurich and other institutions have introduced a new equilibrium concept called the strategically robust Wardrop equilibrium. This concept addresses uncertainty in multiagent systems like electric vehicle (EV) charging and traffic routing, where agents must make decisions without full knowledge of others' behavior. The approach uses an optimal-transport-based ambiguity set centered on the emergent aggregate population behavior, allowing each agent to protect against worst-case deviations. This framework enables interpolation between standard Wardrop equilibria (no robustness) and security strategies (maximum robustness), providing a flexible tool for modeling real-world scenarios.

The team proved the existence of pure strategically robust Wardrop equilibria in convex aggregative games and developed tractable computational methods for solving them. In EV charging simulations, the new equilibrium reduced costs for all agents compared to standard approaches, uncovering a 'coordination-via-robustification' effect where individual robustness leads to collective benefits. This work, published on arXiv (2604.23669), bridges game theory and optimization, with applications in smart grid management, autonomous vehicle coordination, and other domains where agents face uncertain aggregate behavior from limited computation or bounded rationality.