A typical teleoperation system consists of a master system, slave system, and a communication channel connecting the master and slave. For stable teleoperation, the slave end must track the motion data sent by the teleoperator at the master side accurately. And the interaction force between the slave and the remote environment should be rendered to the operator accurately. In Model-Mediated Teleoperation, the slave environment is modeled approximately at the master end and force is rendered from this local model.
Many Model-Mediated Teleoperation systems present in the literature treat the interaction between telerobot and environment as an interaction at a specific point. However, dexterous interaction involves the use of hands which interacts with the environment at several points. Human beings feel sensations at different parts of their hand while manipulating objects. The gripping sensation is felt by fingers and palm, while the object manipulation is more dominant at the wrist. The gripping sensation is also felt during manipulation. The hypothesis for this research work is that an object’s dynamics can be broken down into two different dynamic sub-models: one for approximating the gripping force and the other for approximating the forces felt due to object manipulation. The gripping/contact dynamics model is called the flexible dynamics model or the flexible model. And the object manipulation model is called the inertial dynamics model or inertial model. Due to splitting an object’s dynamics, there can be performance issues in model-specific estimations. One sub-model can introduce parasitic effects in the other and/or vise-versa.
The goals of this assignment were to design inertial and flexible models for an object. This is followed by designing estimation techniques for estimating the inertial and flexible model parameters. Finally, validating this method to prove or disprove the hypothesis. The results show good performance at high frequencies for the inertial model. The flexible model suffers from inaccuracies throughout the experiment. And the inertial model suffers from inconsistencies at low frequencies. It was proven that the flexibility of the object is not the reason for inconsistencies at low frequencies for the inertial model. The flexible model is not descriptive enough. In the future, different models for flexible dynamics can be tested which would yield better results.
To join the presentation via Microsoft Teams click here