Integration of visual and joint information to enable linear reaching motions

Eberle, H., Nasuto, S. ORCID: https://orcid.org/0000-0001-9414-9049 and Hayashi, Y. ORCID: https://orcid.org/0000-0002-9207-6322 (2017) Integration of visual and joint information to enable linear reaching motions. Scientific Reports, 7. 40869. ISSN 2045-2322 doi: 10.1038/srep40869

Abstract/Summary

A new dynamics-driven control law was developed for a robot arm, based on the feedback control law which uses the linear transformation directly from work space to joint space. This was validated using a simulation of a two-joint planar robot arm and an optimisation algorithm was used to find the optimum matrix to generate straight trajectories of the end-effector in the work space. We found that this linear matrix can be decomposed into the rotation matrix representing the orientation of the goal direction and the joint relation matrix (MJRM) representing the joint response to errors in the Cartesian work space. The decomposition of the linear matrix indicates the separation of path planning in terms of the direction of the reaching motion and the synergies of joint coordination. Once the MJRM is numerically obtained, the feedfoward planning of reaching direction allows us to provide asymptotically stable, linear trajectories in the entire work space through rotational transformation, completely avoiding the use of inverse kinematics. Our dynamics-driven control law suggests an interesting framework for interpreting human reaching motion control alternative to the dominant inverse method based explanations, avoiding expensive computation of the inverse kinematics and the point-to-point control along the desired trajectories.

Item Type	Article
URI	https://reading-clone.eprints-hosting.org/id/eprint/68855
Item Type	Article
Refereed	Yes
Divisions	Life Sciences > School of Biological Sciences > Department of Bio-Engineering
Publisher	Nature Publishing Group
Download/View statistics	View download statistics for this item