Perception of Prototypical Environments in Model-Mediated Teleoperation

Abstract

In teleoperation, master and slave robots directly exchange position and force data to establish a connection between local and remote sites. Model-mediation interjects an environment model rendered to the user to create a sense of remote presence. The slave continually updates this model from actual interactions. Here we evaluate a dual-mode implementation, which classifies the environment into one of two models describing rigid contact or free motion. Performance is measured against various prototypical environments, observing master force outputs to motion inputs to evaluate user’s perception of the environment. We find performance to depend on system lag as well as model error, such that multiple-model systems fundamentally outperform traditional architectures.