Design of a MMT system for an environment with changeable dynamics

Telemanipulation enables a user to operate in a remote environment by using a robotic system. The teleoperated system projects the operator's actions in the remote environment, allowing the application of human dexterous skills in hostile or unreachable environments to the human user.

An ideal telemanipulation system could produce a complete immersion in the remote environment for the users. Meaning that it is able to replicate the user movements on the remote side and the users are able to perceive the remote environment as they were present. The main challenges for Teleoperation are the time delays between the master and slave communication systems and the creation of a control strategy that is able to accomplish the functionality described above.

Model-Mediated Teleoperation (MMT) is an approach in the field. It uses sensor data from the remote environment to construct and update a competent model of its environment. These models should be sufficient to describe the physical aspects of the behaviour of the remote environment, so they are able to replicate the environment in the local system, anticipating interactions for the user as it avoids time delays due to communications. In MMT, the main teleoperation challenge is modelling the environment. For modelling, the system needs to estimate the parameters that comprise the physical behaviour of the model. How precise the estimation of parameters is in a given time makes an MMT system varies between a steady-state or transition state. In the transition state, the remote environment and the local virtual model do not match. During this state, the operator won't be able to perceive the environment correctly. An MMT system should minimize the time of the transition state by accelerating the convergence of the parameter estimation. In the steady-state, a perfect match between the local and the environment model is achieved.

Ideally, the system converges to the steady-state in a static environment when the parameters are stable and there is no external interaction. However, in a dynamic environment, the environment model changes depending on how the user interacts. If the model changes, new parameters are included in the system and an update of the system is required to adapt the system.

The main objective of this graduation project lies in applying the MMT system in a dynamic environment. An MMT system should adapt to the transitions in the model and its dynamics. In the case of a movable object, these transitions are faced when the object changes its interaction with the environment. For example, when it is dragged against a surface, when it is lifted, etc. The transitions accommodate the model to the changes in the environment, by a change in the position of the object, external forces applied to it, or changes in the geometry of the environment.

This assignment aims to design an MMT system that can adequately display the model switches to the user without interrupting its telemanipulation experience. This will involve:
• Design of a dynamic model for a movable object.
• Create a system that can estimate the physical parameters of the dynamic model before entering into contact with the object.
• Integrate the model switching in an MMT approach.