Enabling Effective Information Sharing in Human-Robot Teams

Artificial agents are increasingly interacting with humans to support them in myriad tasks [21, 29, 30]. Gradually, the role of robots in these interactions is transitioning from that of a passive tool for the human user to that of a proactive collaborator. While noteworthy progress has been made in recent years [26], achieving seamless collaboration between robots and humans remains an active and challenging problem. Realizing the full benefit of human-robot collaboration hinges on the resolution of several challenges, including the development of algorithms that enable robots to (a) model and predict the behavior of humans, and (b) utilize these models to make execution-time decisions. In several domains, such as disaster response and collaborative manufacturing, these algorithmic challenges are further exacerbated due to the adaptive and inter-dependent nature of human and machine behavior, and partial knowledge of the environment that the human-robot team is operating in. We believe that effective sharing of information between the human and the robot is key to successful resolution of these challenges and achievement of seamless human-robot collaboration. This belief is founded in prior research on human teams [6, 8, 10] and human-machine teams [12, 14, 22], which indicate that effective communication is critical to the success of the collaboration. Thus, we are developing algorithms for human-machine collaboration that (a) leverage communication between humans and robots to learn models of human’s decision-making and (b) enable robots to effectively share information during execution of human-robot collaborative tasks.