@inproceedings {pub4014,
	title = {Online Learning of Feed-Forward Models for Variable Impedance Control in Manipulation Tasks},
	author = {Michael  J. Mathew AND Saif Sidhik AND Mohan Sridharan AND Morteza Azad AND Akinobu Hayashi AND Jeremy Wyatt},
	year = {2019},
	month = {June},
	abstract = {While performing a new manipulation task, humans tend to be stiffer in the initial trials to improve task accuracy and to counter any disturbances that may affect task performance. After a sufficient number of repetitions, humans are able to perform the task with lower stiffness without causing any significant reduction in task performance. Existing literature in human and animal motor control indicates that learned internal models of the manipulation task and the environment are used to predict the state in response to particular control actions, adapting stiffness on the fly to minimise energy usage. Inspired by these findings, we describe a framework for robot manipulation that makes a significant departure from existing work. This framework learns a state-dependent forward model of any given manipulation task and supports hybrid force-motion control. Errors in the predictions of the forward model are used to dynamically adapt the impedance parameters of the feedback controller, and the hybrid controller supports compliance in certain directions. We evaluate the capabilities of our framework in the context of continuous-contact manipulation tasks involving varying external forces. },
	publisher = {workshop webpage (no copyright transferred)},
	booktitle = {Robotic Science and System: Workshop on Task-Informed Grasping},
	address = {https://lcas.lincoln.ac.uk/wp/tig-ii/}
}