Model-mediated teleoperation for movable objects: dynamics modeling and packet rate reduction

We study Model-Mediated Teleoperation (MMT) for objects which can be moved and rotated on a planar surface. In MMT, a simple object model is employed on the master side to approximate the remote environment. The haptic feedback is rendered locally on the master side based on the model whose parameters are estimated on the slave side. Since the haptic control loop is running locally without noticeable delay, system stability can be guaranteed in the presence of arbitrary communication delays. The main challenges for MMT are environment modeling and online parameter identification. This work has two contributions. Firstly, we describe how to model movable objects with approximately known model geometry for MMT systems. The object dynamics including environment damping as well as the friction are modeled in real time during teleoperation. Secondly, we propose a hybrid deadband-based and time-triggered updating scheme to adaptively reduce the required packet rate while transmitting the estimated model parameters from the slave to the master. According to our experiments, the movable object can be accurately modeled in real time. In addition, compared to the conventional deadband-based updating scheme, our proposed hybrid updating scheme shows improvements in both model accuracy and packet rate reduction.