Enabling autonomous information-gathering and self-recovery behaviors in partially-known, communication-constrained environments
Jason M. Gregory, Jeffrey N. Twigg, Jonathan R. Fink · 2016
We present a solution to the problem of autonomously identifying a safe region in a partially-known environment from which a mobile robot will navigate to and transmit mission-critical data to an operator. This work is in the context of humanitarian-assistance, disaster-relief missions where an operator tasks the robot from large standoff distances through extreme, real-world environments and robust operations are required to ensure optimized information gathering. Our proposed approach, referred to as the informed recovery behavior, identifies and selects from candidate regions by modeling both navigation and communication costs using experience-based samples. Finally, we experimentally validate the feasibility of our method and demonstrate an improvement in reducing the time required to navigate and transmit data to an operator, compared to a realistic, baseline approach, through a series of trials in an uncontrolled, real-world environment.