Towards Heterogeneous Multi-Agent Systems in Space

As people venture into space, they will bring with them fleets of robots. Maintaining people's trust in their robot teammates needs to be studied in a way that is realistic about robots being less than completely reliable. It has been shown in the past that when one agent in a system is unreliable, people tend to stop using the whole system [11]. To mitigate this, we created a single identity shared among all the agents in a multi-agent system. We conducted a study where participants worked to make repairs on a lunar base with robots that made several errors. Results are still being collected, but so far the data indicates that a shared identity mental model tends to be better at maintaining system-wide trust than a model where the agents had separate identities. In my future work, I wish to investigate how this shared identity could be used to make the loss of robots in space (like the Opportunity Rover) less distressing to their human teammates, and study methods for visualizing shared identity and its transfer through a system of machine agents.