Shenoy Prajwal, Varadhan S K M
Department of Mechatronics, Manipal Institute of Technology, Manipal Academy of Higher Education (MAHE), Manipal, Karnataka, 576104, India.
Department of Applied Mechanics and Biomedical Engineering, Indian Institute of Technology Madras, Chennai, Tamil Nadu, 600036, India.
Sci Rep. 2025 May 23;15(1):17943. doi: 10.1038/s41598-025-02680-x.
The human hand is a complex manipulator with many joints that can perform various tasks. Neuroscience research has demonstrated that to perform any posture, the brain does not control the individual joints but relies on coactivation patterns called synergies that simultaneously control a set of joints. A combination of these synergies can then be used to reconstruct a variety of postures. While such a hypothesis has been demonstrated for single-handed tasks, a question that is not well-explored is whether such synergies can simultaneously control the joints of both hands during bimanual tasks. This paper attempted to address this question by exploring synergies obtained by performing Principal Component Analysis (PCA) on the kinematic data recorded from both the dominant and non-dominant hands of the participants as they performed bimanual tasks. The ability of synergies to reconstruct postures from a lower-dimensional subspace was presented, and an analysis of the separability of postures was performed using a classification algorithm. The results showed that the first 3 synergies explained greater than 80% variance in data, indicating that a few bimanual synergies can be utilized to control the fingers of both hands. The first three synergies could reconstruct postures with a Root Mean Square Error (RMSE) of 4° and classify tasks with an accuracy of 90%, demonstrating that the task-related information was retained in the lower dimensional subspace. This could significantly reduce control complexities while designing robotic or prosthetic distal upper limb devices.
人手是一个复杂的操纵器，有许多关节，能够执行各种任务。神经科学研究表明，为了做出任何姿势，大脑并非控制单个关节，而是依赖于称为协同作用的共同激活模式，这些模式同时控制一组关节。这些协同作用的组合随后可用于重构各种姿势。虽然这种假设已在单手任务中得到证实，但一个尚未得到充分探索的问题是，在双手任务中，这种协同作用能否同时控制双手的关节。本文试图通过探索对参与者在执行双手任务时从优势手和非优势手记录的运动学数据进行主成分分析（PCA）所获得的协同作用来解决这个问题。展示了协同作用从低维子空间重构姿势的能力，并使用分类算法对姿势的可分离性进行了分析。结果表明，前3种协同作用解释了数据中超过80%的方差，这表明可以利用一些双手协同作用来控制双手的手指。前三种协同作用能够以4°的均方根误差（RMSE）重构姿势，并以90%的准确率对任务进行分类，这表明与任务相关的信息保留在了低维子空间中。这在设计机器人或假肢远端上肢设备时可显著降低控制复杂性。
