Predictive bilateral master-slave manipulation with statistical environment model

A bilateral master-slave control algorithm is proposed. It incorporates a task-environment impedance-model and its modelling error to improve the performance in the presence of time delay between master- and slave-manipulator. The model at each point in slave-manipulator task-space is updated statistically by comparing real sensory information with simulation data. In the model, each object is given stiffness beyond its estimated surface within the range of its modeling error. The technique limits the contact velocity with the real object, thus reducing instability, while maintaining ease of operation elsewhere. The algorithm is experimentally validated