以動態最佳化方法結合慣性量測估測人體運動狀態與地面反作用力:反向跳行為分析

Abstract

本研究提出一個僅使用四顆慣性量測單元 (Inertial Measurement Units, IMU) 估測人體蹲跳動作中運動狀態、地面反作用力及肢段力矩的方法。傳統上，這類分析通常需要使用較昂貴的設備，如測力板，並且在實驗室環境中進行。然而，在非實驗室環境中進行人體運動量測，對於遠端居家復健和多元化運動項目之分析至關重要。在蹲跳行為中，地面反作用力位置對於人體運動狀態的計算結果影響很大，但多數論文只有討論利用 IMU 量測地面反作用力之大小而非位置。 本研究提出的方法主要基於最佳控制和感測器融合策略。首先建立一個人體肢段的動態系統，以各肢段力矩做為系統輸入，輸出為符合 運動學約束的運動狀態。透過最佳化方法求得最佳之系統輸入，在確保地面反作用力位置合理的前提下，使系統輸出與多個感測器測量結果相符合。研究結果表明，此方法能夠在考慮地面反作用力位置合理性的情況下，得到與直接 IMU 測量相似的運動學數據，同時滿足平面簡化模型中的運動學約束。 總的來說，本研究為非實驗室環境下的人體運動分析提供了一種新的透過動態最佳化建立人體運動系統的方法，對於遠端復健、運動訓 練等領域具有重要的應用價值。

This study proposes a method to estimate the human motion, ground reaction force (GRF), and segmental torques during human countermovement jumps using only four inertial measurement units (IMUs). Traditionally, such analyses require more expensive equipment, such as force plates, and are conducted in laboratory settings. However, measuring human motion in non-laboratory environment is crucial for remote rehabilitation and the analysis of diverse sports activities. In countermovement jump behavior, the position of the ground reaction force significantly impacts the calculation results of the human motion state, yet most studies only discuss measuring the magnitude of the ground reaction force using IMUs, rather than its position. The method proposed in this study is primarily based on optimal control and sensor fusion strategies. A dynamic system of human segments is established, using segmental torques as system inputs and outputs that conform to kinematic constraints as the system’s outputs. The optimal system inputs are determined through optimization methods, ensuring that the ground reaction force position is reasonable and that the system outputs match the measurements from multiple sensors. The results indicate that this method can yield kinematic data similar to those obtained from direct IMU measurements while considering the reasonableness of the ground reaction force position and sat- isfying the kinematic constraints of a simplified planar model. In summary, this study provides a novel method for establishing a human motion system through dynamic optimization for human motion analysis outside of laboratory environments, which has significant application value in fields such as remote rehabilitation and sports training.