Recent works in intelligent autonomy for unmanned naval vehicles emphasizes the role and limitations of the human-in-the loop when a single human tries to supervise a team of distal UxVs. In particular, the “fan-out” problem identifies fundamental scaling limits of this approach governed by the cognitive, attentive, and communication burden placed on the human operator. Our scientific goal is to achieve a dramatic improvement in the flexibility, robustness, and scalability of human-robot teams in dynamic and uncertain environments. Our technical goals are to develop the underlying principles, methodologies, and technologies to enable autonomous robot teammates to collaborate naturally and effectively with humans as peers, whether working remotely from their human teammates or side-by-side. Furthermore, the robots must behave compatibly with human cognitive and communicative abilities and limitations. Our evaluation scenario is an urban human-robot teaming task (CBRNE rescue scenario and search and retrieval tasks) where a response team consists of a small number of humans and a much larger number of autonomous ground and aerial robots. The robots interact with human teammates and bystanders (human victims).
J. How, N. Roy (MIT), P. Hinds (Stanford), J. Adams (Vanderbilt), R. Grupen (UMASS Amherst), D. Fox (Univ. Washington)
ONR MURI Grant – Cognitively Compatible and Balanced Human-Robot Teaming in Urban Military Domains (award number N000140710749)