运动学
惯性测量装置
地面反作用力
运动捕捉
计算机科学
步态
工作流程
模拟
部队平台
人工智能
运动(物理)
物理医学与康复
医学
物理
经典力学
数据库
作者
Luana Bottini,Giacomo Di Raimondo,Bryce A. Killen,Z. Sawacha,Ilse Jonkers
标识
DOI:10.1016/j.gaitpost.2023.07.061
摘要
Musculoskeletal modeling (MSKm) and simulation provide an ideal framework for estimating movement-related parameters such as joint contact forces which cannot be measured in vivo. However, these parameters are not well-documented in free-living situations as inertial measurement units (IMUs) have rarely been combined with MSKm workflows. Although IMU-based systems are valuable alternatives to camera-based systems to estimate kinematics, no robust and validated method exists to estimate ground reaction forces (GRF) thereby alleviating the need for force plates. The aim of this study was to implement and validate a workflow for GRF estimation using a combined IMU and MSKm workflow. The ability to estimate such information, when combined with MSKm would provide clinicians access to movement-related relevant to the onset/progression of musculoskeletal diseases in an ecological environment. Can IMU-based kinematics combined with an MSKm-based simulation workflow be used to accurately estimate GRF during gait? Two healthy subjects (1 man, 1 female; age: 27.5±2.2 y) walked overground (self-selected speed). MotionCapture (100 Hz) and InertialCapture (60 Hz) systems were used to record 3D-marker trajectories and body segment orientations. GRF were recorded using force plates (1000 Hz). MoCap and InCap kinematics were calculated using OpenSim [1] and sensors-to-segment calibration OpenIMUs [2], respectively. Musculoskeletal model [3] containing 35 DOFs, 49 muscles, and 5 foot contact spheres was used in both approaches. OpenSim Moco was used to track the IMU-based kinematics with two main goals: (i) minimize the error between input kinematics and simulated motion, (ii) minimize the sum of the absolute value of the muscle activations. MocoSolver was applied to compute the optimized motion and muscle activations. GRF were evaluated based on the generated forces of the foot contact spheres. Centre of pressure (COP) was estimated using a Zero Moment Point (ZMP) method [4]. RMSE of the joint angle between MoCap and InCap was less 5° for all joint angles. RMSE between InCap kinematics and Moco simulated kinematics were 1-6° for all joint angles with knee and ankle flexion/extension max error of 10°. RMSE of anterior/posterior, vertical, and medio/lateral GRF was 0.10±0.02 BW, 0.25±0.01 BW and 0.06±0.03 BW respectively, compared to measured force plate data (Fig. 1). RMSE of anterior/posterior, and medio/lateral COP was 6.7±0.9 cm and 3.8±0.4 cm respectively, compared to measured force plate data.Download : Download high-res image (71KB)Download : Download full-size image The developed workflow for the estimation of GRF based on IMU-based kinematics and OpenSim Moco provided promising results. However, the contact sphere definition strongly affects the estimated GRF with a direct effect on the calculated COP and ultimately joint moments (not shown).
科研通智能强力驱动
Strongly Powered by AbleSci AI