The study of autonomous multi-agent systems has received considerable attention in recent years due to their applicability in model complex processes observed in engineering and biology. This work mainly analyzes strategic situations observed in engineering applications, such as a defense system protecting critical infrastructures against attacks from incoming missiles and interceptors defending an asset against intrusions. These situations are usually analyzed using the mathematical framework of pursuit-evasion games with three players–Target, Attacker, and Defender–and referred to as a Target-Attacker-Defender (TAD) game. This work explores two main themes related to multi-agent TAD games: (i) agents with role-switching behavior; (ii) agents with restricted sensing capabilities.



In the first research problem, we study a multi-player TAD game with switching strategies where the defender is equipped with flexible (and powerful) capability to autonomously change roles from being a rescuer to an interceptor and vice-versa. Augmenting the receding horizon approach with the open-loop Nash equilibrium solution concept, geometric characterization of the trajectories and associated behaviors of the players are studied.



In the second research problem, we study multi-player TAD interactions where defenders have limited sensing capabilities. The visibility constraints of the players induce a visibility network that encapsulates visibility information during the evolution of the game. We obtain network-adapted feedback Nash equilibrium strategies using an inverse game theory approach. We present a consistency criterion for the refinement of these strategies and provide an optimization-based method for computing them.