Accepted author manuscript, 2.09 MB, PDF document
Available under license: CC BY
Final published version
Research output: Contribution to Journal/Magazine › Journal article › peer-review
Research output: Contribution to Journal/Magazine › Journal article › peer-review
}
TY - JOUR
T1 - Learning to Assist Bimanual Teleoperation using Interval Type-2 Polynomial Fuzzy Inference
AU - Wang, Ziwei
AU - Fei, Haolin
AU - Huang, Yanpei
AU - Rouxel, Quentin
AU - Xiao, Bo
AU - Li, Zhibin
AU - Burdet, Etienne
PY - 2023/5/3
Y1 - 2023/5/3
N2 - Assisting humans in collaborative tasks is a promising application for robots, however effective assistance remains challenging. In this paper, we propose a method for providing intuitive robotic assistance based on learning from human natural limb coordination. To encode coupling between multiple-limb motions, we use a novel interval type-2 (IT2) polynomial fuzzy inference for modeling trajectory adaptation. The associated polynomial coefficients are estimated using a modified recursive least-square with a dynamic forgetting factor. We propose to employ a Gaussian process to produce robust human motion predictions, and thus address the uncertainty and measurement noise of the system caused by interactive environments. Experimental results on two types of interaction tasks demonstrate the effectiveness of this approach, which achieves high accuracy in predicting assistive limb motion and enables humans to perform bimanual tasks using only one limb.
AB - Assisting humans in collaborative tasks is a promising application for robots, however effective assistance remains challenging. In this paper, we propose a method for providing intuitive robotic assistance based on learning from human natural limb coordination. To encode coupling between multiple-limb motions, we use a novel interval type-2 (IT2) polynomial fuzzy inference for modeling trajectory adaptation. The associated polynomial coefficients are estimated using a modified recursive least-square with a dynamic forgetting factor. We propose to employ a Gaussian process to produce robust human motion predictions, and thus address the uncertainty and measurement noise of the system caused by interactive environments. Experimental results on two types of interaction tasks demonstrate the effectiveness of this approach, which achieves high accuracy in predicting assistive limb motion and enables humans to perform bimanual tasks using only one limb.
KW - Artificial Intelligence
KW - Software
U2 - 10.1109/tcds.2023.3272730
DO - 10.1109/tcds.2023.3272730
M3 - Journal article
SP - 1
EP - 1
JO - IEEE Transactions on Cognitive and Developmental Systems
JF - IEEE Transactions on Cognitive and Developmental Systems
SN - 2379-8920
ER -