Hand Posture Reconstruction Through Task-Dependent Hand Synergies

Bingchen Liu,Li Jiang,Shaowei Fan
DOI: https://doi.org/10.1007/978-3-030-89095-7_2
2021-01-01
Abstract:Currently, the propose of postural synergies theory has made great contribution to analysis of human hand function. However, most studies ignored the hand synergy patterns varying in different kinds of grasping tasks. Therefore, a set of task-dependent hand synergies was proposed in this paper. Firstly, a functional hand movement dataset was established, and “The GRASP Taxonomy” proposed by Thomas Feix was chosen as the experimental paradigm. Then, in order to extract specific coordination strategy in similar tasks, these 33 kinds of grasp types were clustered into six groups. Next, Principal Components Analysis (PCA) was separately applied on each cluster of tasks instead of the whole dataset to get task-dependent hand synergies. Finally, the postural reconstruction error was used as evaluation indicator for comparing the accuracy of synergy model proposed in this paper and previous invariant synergy model. The results showed that the mean joint reconstruction error in this improved model has been reduced by 63.98% using the same number of hand synergies, verifying the necessity of adjusting synergy strategy according to task requirement.
What problem does this paper attempt to address?