目的 针对下肢外骨骼结构优化,控制算法有效性及穿戴舒适性评估的研究,客观上需要对人机交互动力学及穿戴者肌骨系统的动态负荷等特征进行量化分析。然而,目前采用力学传感技术对人与外骨骼间的六维交互力进行直接测量仍是一个挑战性课题。方法 本研究提出一种基于虚拟肌肉的人机耦合动力学建模方法。首先在外骨骼穿戴实验中,利用人体动作捕捉系统和自行开发的力学监测装置,同步获取穿戴者的步行动力学、肌电信号、外骨骼驱动状态及局部的法向人机交互力信息；然后,在肌骨系统建模环境中建立人机耦合模型,并以步态实验数据和外骨骼关节扭矩作为耦合模型的驱动信息,进行逆动力学计算；最后,通过调整虚拟肌肉的强度以及刚度参数,对模型的仿真数据与实验测试结果进行比对,量化评估下肢外骨骼人机耦合模型的有效性。结果 结果表明：耦合模型逆动力学计算的法向交互力以及下肢肌肉激活情况与步态实验测量结果相比,在响应曲线趋势上均具有良好一致性,法向交互力在步态周期中均出现了5个峰值,X正方向交互力的第一峰值误差为8%,第二和第三峰值误差分别为23%和22%,X负方向的两个交互力峰值误差分别为0.1%和2%,下肢肌肉激活程度峰值误差均小于5%。结论 本文提出的人机耦合模型可有效计算人与外骨骼交互力。该耦合模型的建立,能为外骨骼结构优化与控制算法的验证与迭代,以及外骨骼助行助力功效的性能评估提供理论依据。
Objective In view of the optimization of lower extremity exoskeleton structure, the effectiveness of control algorithms and the evaluation of wearing comfort, it is objectively necessary to quantitatively analyze the characteristics of human-computer interaction dynamics and dynamic load of the musculoskeletal system. However, it is still a challenging subject to directly measure the six-dimensional interaction force between human and exoskeleton using mechanical sensing technology. Mtehods This research proposes a human-machine coupling dynamics modeling method based on virtual muscles. First, in the exoskeleton wear experiment, the human motion capture system and self-developed mechanical monitoring device are used to obtain the wearer’s walking dynamics, electromyography signals, exoskeleton drive status and local human-computer interaction information; The human-machine coupling model is established in the bone system modeling environment, and the gait experiment data and the exoskeleton joint torque are used as the driving information of the coupling model to perform inverse mechanical calculations. Finally, by adjusting the strength and stiffness parameters of the virtual muscles, the real data of the model is compared with the experimental test results to quantitatively evaluate the effectiveness of the human-machine replacement model of the lower extremity exoskeleton. Results The results show that the normal interaction force calculated by the inverse dynamics of the coupled model and the activation of the lower limb muscles have good consistency in the response curve trend compared with the measurement results of the gait experiment. The normal interaction force has 5 peaks in the gait cycle. The error of the first peak of the interaction force in the positive X direction is 8%, the error of the second and third peaks are 23% and 22%, and the error of the peak of the two interaction forces in the negative X direction is 0.1% and 2%, respectively , The peak error of lower limb muscle activation is less than 5%. Conclusions The human-machine coupling model proposed in this paper can effectively calculate the interaction force between human and exoskeleton. The establishment of the coupling model can provide a theoretical basis for the verification and iteration of the exoskeleton structure optimization and control algorithm, as well as the performance evaluation of the exoskeleton"s mobility assistance effect.