Collaborators: Xuan Wang (GMU), Xuesu Xiao (GMU)

We are interested in designing coordinated group motion, where the safety or cost for one agent to move between locations may depend on the support provided by its teammate. 

Specifically, we are interested in a scenario where a team of robots cooperatively traverse a challenging environment by "supporting'' each other. Support can take the form of, for example, providing a different vantage point for better situational awareness, or physically interacting with the environment to reduce risk (e.g., holding a ladder). We abstract these notions to the actions that can be taken on a graph environment to study how multi-agent teams can efficiently traverse an environment when such cooperation is possible.

Students: James Berneburg, Manshi Limbu, Sara Oughourli

Collaborators: Scott Guan (GaTech), Austin Chen (UPenn), Panagiotis Tsiotras (GaTech), Vijay Kumar (UPenn), Jason Marden (UCSB)

Deploying resources (robots, sensors, or supplies) to appropriate locations at the appropriate time is a fundamental problem in multi-agent systems.  In this work, we study the dynamic resource allocation problem on a graph, where nodes represent physical locations and edges represent the traversability between those locations. The focus is on transporting the resources effectively in the environment to satisfy demands that change dynamically. 

To stress the dynamic aspect of the problem, we consider demands that are generated by an adversary.  Specifically, we formulate the problem as a dynamic (turn-based) game played between a blue team of defender robots and a red team of attacker robots.  The defender team must ensure numerical advantage at every node where the attacker robots are present. By solving this game, we study strategic and dynamic allocation of resources.