A Panoramic Model for Remote Robot Environment Mapping and Predictive Display

Abstract

Robot tele-operation is significantly degraded by delays in the operator visual feedback. We present a cylindrical image-depth model that can be automatically acquired at the remote work site and then used to support both robot localization and predictive display. We present an implementation for a mobile robotics system where synthesized immediate visual feedback replaces the delayed real images. Experimentally we found that a predicted view could be rendered with an geometric viewpoint error of no more than 2.4cm/1.2 degrees in a 100m 2 work scene.