STRUCTURE PRESERVING OPTIMAL CONTROL OF FINGER MOVEMENTS
STRUCTURE PRESERVING OPTIMAL CONTROL OF FINGER MOVEMENTS
复制标题
保持手指运动最佳控制的结构
DOI：
--
复制
发表时间：
2011
期刊：
影响因子：
--
通讯作者：
S. Leyendecker
中科院分区：
文献类型：
--
作者：
R. Maas;T. Siebert;S. Leyendecker
解决动力学问题的常用工具，特别是在生物力学研究中，是MATLAB/Simulink。许多积分方法，例如在MATLAB/Simulink中使用的，依赖于连续运动方程的标准离散化。这些方法通常导致时间步进格式，显示能量和动量的数值耗散。与此相反，我们使用离散变分原理推导出一个时间步进格式。这种方法产生离散的类似物的欧拉拉格朗日方程和诺特定理，这确保了底层的连续动力系统的结构被保存。利用这种方法，模拟结果是辛动量一致的，并表现出良好的能量行为。我们实现了一个典型的非线性Hill型肌肉模型的结构保持仿真框架，并研究与标准的肌肉驱动运动的MATLAB/Simulink仿真的差异，特别是关于能量和角动量的正确表示。数值算例表明，MATLAB/Simulink积分器人为地释放或获得能量和角动量，而辛动量法的所有结果都是结构保持的。与此结构保持仿真框架，包括驱动的肌肉模型，我们调查的手指在抓取运动的轨迹。由于人体运动是由中枢神经系统（CNS）控制的，我们将手指运动公式化为具有生理动机目标函数的约束强迫运动的最优控制问题，如[1]所述。对于最优控制问题的解决方案，我们使用DMOCC（离散力学和约束系统的最优控制，在[2]中介绍），它可以通过其结构保持公式与其他直接转录方法区分开来。这是该方法的一个关键特征，因为数值耗散可能导致高估或低估关节扭矩或肌肉力。
A common tool to solve dynamical problems, in particular in biomechanic investigations, is MATLAB/Simulink. Many integration methods, as used for example in MATLAB/Simulink, rely on standard discretisations of the continuous equations of motion. These methods often lead to time stepping schemes, that show numerical dissipation in energy and momentum. In contrast to that, we use a discrete variational principle to derive a time-stepping scheme. This method yields discrete analogues to the EulerLagrange equations and Noether’s theorem, which ensures that the structure of the underlying continuous dynamical system is preserved. Using this method, the simulation results are symplectic momentum consistent and exhibit a good energy behaviour. We implement a typical nonlinear Hill-type muscle model in the structure preserving simulation framework and investigate the differences to standard simulation of muscle actuated movements with MATLAB/Simulink, especially concerning the correct representation of energy and angular momentum. A numerical example shows that the MATLAB/Simulink integrators artificially loose or gain energy and angular momentum, whereas all results of the symplectic momentum method are structure preserving. With this structure preserving simulation framework including actuation by muscle models, we investigate the trajectory of fingers during grasping movements. Since human movements are controlled by the central nervous system (CNS), we formulate finger movements as optimal control problems for constrained forced motion with a physiologically motivated objective function as described in [1]. For the solution of the optimal control problem, we use DMOCC (Discrete Mechanics and Optimal Control for Constrained Systems, introduced in [2]), which can be distinguished from other direct transcription methods by its structure preserving formulation. This is a key feature of the method, since numerical dissipation could lead to overor underestimation of the joint torques or muscle forces.
影响因子：
1.8
作者：
S. Leyendecker;S. Ober-Blöbaum;J. Marsden;Magdalena Ortiz
通讯作者：
S. Leyendecker;S. Ober-Blöbaum;J. Marsden;Magdalena Ortiz
影响因子：
--
作者：
F. Zajac
通讯作者：
F. Zajac
影响因子：
2.5
作者：
Kamper, DG;Cruz, EG;Siegel, MP
通讯作者：
Siegel, MP