The main subject of this thesis is on interaction control cast into an impedance control framework. Literature does not provide a unified framework to define the impedance values on the basis of a task. Recent developments in data-driven approaches (like machine learning) treat the system as a black box and learning impedance profiles directly from sensor feedback. However, many impedance controlled robots operate in more structured environments.
Therefore, the core idea put forward in this work is that the knowledge about the task, robot dynamics and its environment can be utilised to define 'good' initial values for varying impedance that allow completion of the task in nominal conditions. We define a control strategy as the output of an optimization based on time-varying Cartesian impedance values which results in an open-loop control action. This feedforward will be chosen on the basis of the task. The resulting task-based feedforward is supplemented by a task-free impedance controller to handle environmental uncertainties and external perturbations.
The thesis validates the efficacy of the proposed methods by presenting simulation studies of a simple mass and a 5-DoF robot.
BlueJeans videoconference join information:
Meeting URL
https://bluejeans.com/767678754
Meeting ID
767 678 754
Want to dial in from a phone?
Dial one of the following numbers:
+31.20.808.2256 (Netherlands (Amsterdam))
(see all numbers - https://www.bluejeans.com/numbers)
Enter the meeting ID and passcode followed by #