Detection of discrepancies and sensory-based recovery for virtual reality based telemanipulation systems

Teleoperators using an intermediate functional representation of the remote real environment (RE), suffer the lack of accurate synthetic modeling. Discrepancies will always occur between the RE and its artificial representation by means of virtual environment (VE). A strategy to deal with VE/RE uncertainties based mainly on sensory level interpretation is presented. It is directed to avoid the use of task knowledge by providing a stream of virtual sensors values. Within the remote site, a supervisor is in charge of recovering from the VE/RE discrepancies by a continuous simulated and real states comparison. The states are derived from simulated and real sensor interpretations respectively. A simple experiment with the proposed approach is presented using only position (velocity) and force sensors. Limitations of the proposed approach are also discussed