Prediction of Human Gait Adaptations Using Optimization principles

人类通常以可重复的方式执行动作，并根据任务的要求和肌肉骨骼系统施加的限制调整自己的动作。例如，假肢将导致不正常的步态，这不是设备的直接结果，而是身体未受影响部分肌肉控制的适应结果。这项研究检验了这样一种假设，即这种适应是由一般的最优化原则支配的，因此可以预测。肌肉骨骼建模和优化将用于预测人类步态对十种不同条件的适应：五种速度、两种上坡、两种下坡和倒退行走。理论预测将使用四个最优化原则：最小疲劳、最小代谢能量、最小肌肉力量、最小感知努力。这些预测将与12名人类受试者的相应反应进行比较，以检验其有效性，并期望最小代谢能原理将产生最佳预测。这项研究的更广泛的影响将包括它对人类运动的运动控制、生物力学和生物能量学的统一做出的贡献。这些发现将适用于康复设备、假肢、外科干预、人工操作的机器和运动器材的设计。计算技术将通过交互式网络界面向目标受众高中生和本科生公开。参观者将能够定义任务条件和优化标准，并使用建模引擎来预测人类行为，这将使用动画可视化。

0302259van den BogertHumans typically perform movements in a repeatable manner, and adjust their movement to requirements of the task and to constraints imposed by the musculoskeletal system. For instance, a prosthetic limb will lead to an abnormal gait that is not a direct consequence of the device, but is rather the result of adaptations in the control of muscles in the unaffected part of the body. This research tests the hypothesis that such adaptations are governed by a general optimization principle, and can therefore be predicted. Musculoskeletal modeling and optimization will be used to predict the adaptation of human gait to ten different conditions: five speeds, two uphill slopes, two downhill slopes, and walking backwards. Theoretical predictions will be made using four optimization principles: minimal fatigue, minimal metabolic energy, minimal muscle force, minimal perceived effort. The predictions will be compared to the corresponding responses in twelve human subjects to test their validity, with expectations that the principle of minimal metabolic energy will produce the best predictions. A broader impact of this research will include its contribution to a unification of motor control, biomechanics, and bioenergetics for human locomotion. The findings will be applicable to the design of rehabilitation devices, prosthetics, surgical interventions, human-operated machines, and sport equipment. The computational techniques will be made publicly available through an interactive web interface for a target audience of high school and undergraduate students. Visitors will be able to define task conditions and optimization criterion and use the modeling engine to predict human behavior, which will be visualized using animation.