基于自适应CPG的人体节律运动协同特征刻画

人体柔顺自然的节律运动是全身各关节协同有序转动的展现，关节间的耦合时序与动态转动特性内蕴着人体节律运动中各肢体间的协同关系。招募20名年轻健康被试(男女各半)进行行走与跳绳实验，采集运动时主要关节的角度数据；引入自适应Hopf振荡器参数辨识模型并结合关节协同相位分布，针对人体节律运动协同特征的刻画问题进行研究。通过设立关节相位基准点，解算人体关节间耦合转动时序，基于自适应Hopf振荡器，建立关节单元参数辨识模型，获取复杂关节动态转动规律的刻画参数，进而利用中枢模式发生器(CPG)网络的相位耦合特性，重构完整的人体节律协同运动，并对重构结果的准确性进行量化分析。结果表明，基于刻画参数还原的人体节律运动姿态规范，关节重构轨迹与实际数据变化规律高度一致，两者间的相关系数高于0.99,最大平均误差低于0.01 rad,最大误差小于0.03 rad,阈值绝对偏差在4%以下。所提出的关节转动时序计算准则和自适应关节单元参数辨识方法，可准确刻画人体节律运动中的关节耦合特性与协同规律。

Characterization of Human Rhythmic Movement Synergy Based on Adaptive CPG

The human body's natural rhythmic movement is the result of the synergistic and orderly rotation of the joints throughout the body. The coupled timing and dynamic rotational characteristics of the joints embody the synergistic relationship between the limbs in the body's rhythmic movement. In this study, we organized twenty young healthy subjects (10 men and 10 women, 20~26 years old) to perform walking and rope skipping experiments to collect data on the angles of the major joints during exercise, and we introduced the adaptive Hopf oscillator parameter recognition model and combined with the joint synergistic phase distribution to study the problem of portraying the synergistic characteristics of human rhythmic movements. First, by setting up the joint phase reference points, we calculated the coupled rotation timing between human joints. Next, we established a joint unit parameter identification model based on the adaptive Hopf oscillator to obtain the characterizing parameters of the complex joint dynamic rotation patterns. Finally, we used the phase coupling characteristic of central pattern generator network to reconstruct the complete human rhythmic synergetic motion and quantitatively analyzed the accuracy of the reconstruction results. The results showed that the human body’s rhythmic motion posture based on the restoration of the characterization parameters was normal, the joint reconstruction trajectory was highly consistent with the actual data, the correlation coefficient between the two was higher than 0.99, the maximum average error was less than 0.01 rad, and the maximum error was less than 0.03 rad, the absolute threshold deviation was less than 4%. Therefore, the joint rotation timing calculation criteria and the adaptive joint unit parameter identification method proposed in this paper can be used to accurately describe the joint coupling characteristics and synergy laws in a human rhythmic motion.