Sentinel: A System for Command and Control of Small Teleoperated Robots

We propose to develop Sentinel, an integrated system for command and control of teams of small teleoperated/autonomous robots. Sentinel will integrate technologies for dynamic camera selection, variable autonomy levels, variable-bandwidth communication, and Monte Carlo Localization using GPS, INS, and manual landmark recognition. The Sentinel localization system will enable operators to increase localization accuracy by identifying landmarks visible through robot cameras. This system will also enable robots to use other robots as landmarks, providing increased localization accuracy for all robots and providing fault-tolerance when GPS is unavailable. During Phase I, we will design the integrated Sentinel teleoperation/autonomy system. We will design this system to enable the operator to dynamically change robot autonomy levels based on the current situational demands. We will also characterize the bandwidth requirements for command and control of multiple robots, including the bandwidth-vs-video-quality tradeoffs involved in simultaneously receiving multiple video streams. Also during Phase I, we will design the Sentinel localization system, including the system architecture, user interface, and Bayesian sensor models. We will implement and demonstrate an initial version of this localization system using GPS and manual landmark recognition on a PackBot UGV.