Deep Fictitious Play-Based Potential Differential Games for Learning Human-Like Interaction at Unsignalized Intersections
Di cosa parla
Si insegna alle auto a comportarsi in incroci senza semafori combinando un modello di interazione basato sui giochi con l'apprendimento da dati di guida reale. Il metodo usa “fictitious play” — cioè ogni veicolo impara a rispondere osservando le azioni passate degli altri — e apprende pesi che catturano stili di guida diversi; gli autori mostrano che il processo converge a risultati stabili.
Cosa permette di osservare
Permette di esplorare se e come approcci ispirati alla teoria dei giochi, addestrati su dati reali, possono produrre comportamenti di guida simili a quelli umani, distinguere stili di guida e garantire convergenza verso esiti prevedibili.
Dalla fonte
Modeling vehicle interactions at unsignalized intersections is a challenging task due to the complexity of the underlying game-theoretic processes. Although prior studies have attempted to capture interactive driving behaviors, most approaches relied solely on game-theoretic formulations and did not leverage naturalistic driving datasets. In this study, we learn human-like interactive driving policies at unsignalized intersections using Deep Fictitious Play. Specifically, we first model vehicle interactions as a Differential Game, which is then reformulated as a Potential Differential Game. The weights in the cost function are learned from the dataset and capture diverse driving styles. We also demonstrate that our framework provides a theoretical guarantee of convergence to a Nash equilibrium. To the best of our knowledge, this is the first study to train interactive driving policies u…