Towards Brain Metrics for Improving Multi-Agent Adaptive Human-Robot Collaboration: A Preliminary Study

When humans work closely together, they can pick up subtle cues from their team members and adapt their behavior appropriately. Humans working closely with robots may also give off cues, but the robots cannot detect these signals and therefore cannot change behavior. In this paper, we focus on heterogeneous multi-human and robot teams. Such scenarios exist frequently in search and rescue operations as well as space missions, where robots perform tasks that are unsafe or even impossible for humans. At the same time, human team members collaborate to make important decisions that influence and direct the robots' work. These decisions often have to be made quickly with high levels of uncertainty, with simultaneous physical and mental demands on the human. In this project, we aim to explore the following questions: Can brain data provide insights that could improve team performance? Could we use these signals to detect when someone is experiencing excessive workload? Could we detect an impact on team performance caused by the robot?