Mediating Time Delayed Teleoperation with User Suggested Models: Implications and Comparative Study

Model mediation allows telerobotic users to manipulate remote environments across several seconds of delay. As the slave robot interacts with its surroundings an approximate model of the environment is continually estimated and sent to the master, where it is visually and haptically displayed to the user. By interacting with this local rendition the user is able to produce appropriate force and motion commands, which are transmitted to the slave. This paper improves upon the scheme by allowing users to provide their own model estimates as suggestions to the system, in effect predicting an environment the slave has yet to detect. A user study is conducted involving master-slave teleoperation under two seconds of round-trip delay, with two goals: to test the utility of the new feature, and to test the utility of model mediation against existing teleoperation methods. Three comparison methods are chosen which can handle large delays: video-only feedback, video with visual slave predictor overlay, and wave variables. Experimental results are presented which effectively demonstrate both the benefits of user suggested models to model mediation as well as the benefits of model mediation over previous methods.