With most tele-operated robots the operator's only feedback is the view from an onboard camera. Live video lets the operator observe the robot's immediate surroundings but does not establish the orientation or whereabouts of the robot in its environment. An additional plot of the robot's trajectory would be helpful for the operator and is sometimes provided, based on GPS. However, indoors where GPS is unavailable, tracking has to rely on dead-reckoning, which is too inaccurate to be useful. Our proposed TelOpTrak method corrects dead-reckoning errors even when only odometry and a low-cost (and thus, high-drift) MEMS-class gyro are available on the robot. TelOpTrak corrects gyro drift by exploiting the structured nature of most buildings, but without having to directly sense building features. This paper explains the TelOpTrak method and provides comprehensive experimental results.
Earlier versions of this paper (Borenstein et al., 2010a), (Borenstein et al., 2010b) were presented at two conferences. The main difference between the earlier conference papers and the present manuscript is that the latter is more comprehensive, more up-to-date, and it presents an entirely new set of experimental results, including results of a live demonstration at the 2010 Robotics Rodeo event at Ft. Benning, USA.