One Operator to Rule Them All: Human-Robot Interaction for Real-World and Analog Subsurface Exploration

Motivated by space exploration and the search for life, NASA JPL's team CoSTAR is taking part in the DARPA Subterranean (SubT) Challenge. The goal of the SubT Challenge is to deploy a team of robots which autonomously explores challenging subsurface worlds such as man made tunnels, urban environments and natural caves. The latter can be used as analog test sites to develop new capabilities for exploration and science tasks in preparation for the exploration of lava tubes on other planetary bodies.

In this work, we present team CoSTARs Human-Robot interaction concept and technologies to overcome the challenges in such risky and potentially hazardous environments.

When deploying a human-robot team in an unknown environment, operations require keeping the human safe at all times. The human should also be aware of the team's progress and overall status. Self-adaptive autonomy is used in CoSTARs system to present the most imminent information to the operator to issue higher level commands, if needed. History has shown that there can be unforeseen events with autonomous systems and thus having a human-in-the-loop provides enormous benefits in terms of in-situ planning and decision making. The system is also designed to work in environments without communication infrastructure, taking delays and dropouts into account with the goal of forwarding as much information about the subterranean environment as possible to the operator.

Real-world testing and simulation results are presented with a focus on the human-robot interaction and self-adaptive autonomy.