Publication

Autonomy Artificial Intelligence Robotics (AAIR)
2010

HEURISTICS-ENHANCED DEAD-RECKONING FOR IMPROVED SITUATION AWARENESS WITH TELE-OPERATED ROBOTS

by Johann Borenstein; Adam Borrell; Russ Miller; David Thomas

Abstract

This paper presents a practical and easy to implement method for tracking the position of tele-operated Unmanned Ground Vehicles (UGVs) inside buildings, where GPS is unavailable. In conventional dead-reckoning systems, which typically use odometry combined with a singleaxis gyro or an Inertial Measurement Unit (IMU), heading errors grow without bound. For that reason, tracking the position of tele-operated UGVs for more than a few minutes becomes unfeasible. Our method, called Heuristics-Enhanced Dead-reckoning (HEDR), overcomes this problem by completely eliminating heading errors at steady state in tele-operated missions of unlimited duration. As a result, HEDR allows the plotting of very accurate trajectories on the Operator Console Unit (OCU). When overlaid over an aerial photo of a building, the real-time trajectory display gives the operator crucial information about position and heading of the UGV relative to the building. This feature offers the operator much improved situation awareness that is not provided by the conventional video feed from onboard cameras.