MOVEMENT CONTROL IN MANIPULATING DYNAMIC OBJECTS

每天，人类都会与挑战他们控制运动稳定性能力的物体互动，比如拿一杯热咖啡。对于患有运动障碍的患者，例如与基底神经节或小脑疾病相关的患者，此类任务可能会显著影响他们执行日常任务的能力。然而，令人惊讶的是，人们对中枢神经系统（CNS）在与动态复杂物体互动时如何控制运动知之甚少。CNS使用物体力学特性的内部表征来操纵简单的刚性物体，也可以学习复杂外力场的内部模型来进行伸展运动。假设被试学习复杂物体动态特性的类似内部表征。我们实验室的初步实验表明，在目标导向的到达任务中，健康受试者调整他们的运动以保持非刚性物体的稳定性。本研究首先将努力确定这种适应是否由于物体动力学内部模型的发展而发生，其次，描述在操纵复杂物体时与保持运动稳定性有关的适应的那些方面。实验将使用一个机器人操纵杆进行，该操纵杆允许精确定义复杂“虚拟物体”的内在动力学特性。在第一个实验中，受试者将被训练在操纵一个动态复杂的虚拟物体时进行点对点的到达运动。用动态物体和刚性物体进行的运动之间的图形误差（量化运动形状轮廓的差异）的比较将用于确定受试者是否正在学习和物体动力学的内部模型，或者只是简单地增加整体肢体刚度或记忆机械手施加的特定力模式。在第二个实验中，受试者将与动态物体进行连续的有节奏的运动。来自非线性动力学的方法将用于量化运动维度（相关维度）和神经肌肉控制系统对内部产生的局部扰动（李雅普诺夫指数）的敏感性。假设运动的维度是由手臂+物体的内在力学决定的，因此不会随着训练而发生显著变化，但中枢神经系统会调整其控制策略，使运动对局部扰动不那么敏感。预计这些实验的结果将有助于更好地理解物体操纵背后的神经肌肉控制过程，并最终有助于为运动康复开发功能逼真的虚拟交互提供更清晰的方向。

Every day, humans interact with objects that challenge their ability to control movement stability, such as carrying a cup of hot coffee. For patients with movement disorders, such as those associated with basal ganglia or cerebellar disease, such tasks can significantly impact their ability to perform everyday tasks. However, surprisingly little is known about how the central nervous system (CNS) controls movement when interacting with dynamically complex objects. The CNS uses internal representations of object mechanical properties to manipulate simple rigid objects, and can also learn internal models of complex external force fields for making reaching movements. It was hypothesized that subjects learn similar internal representations of the dynamical properties of complex objects. Preliminary experiments from our lab indicate that healthy subjects adapt their movements to maintain stability of non-rigid objects during a goal-directed reaching task. The proposed research will first endeavor to determine if this adaptation occurs as the result of the development of an internal model of object dynamics, and second, to characterize those aspects of adaptation that are specifically related to maintaining movement stability when manipulating complex objects. Experiments will be conducted using a robotic manipulandum that allows the intrinsic dynamical properties of complex "virtual objects" to be precisely defined. In the first experiment, subjects will be trained to make point-to-point reaching movements while manipulating a dynamically complex virtual object. A comparison of figural errors (which quantify differences in kinematic shape profiles) between movements made with dynamic and rigid objects will be used to determine if subjects are learning and internal model of the object dynamics, or are simply increasing overall limb stiffness or memorizing the specific patterns of forces imposed by the manipulandum. In the second experiment, subjects will make continuous rhythmic movements with the dynamic object. Methods from nonlinear dynamics will be used to quantify movement dimensionality (correlation dimensions) and the sensitivity of the neuromuscular control system to internally-generated local perturbations (Lyapunov exponents). It is hypothesized that the dimensionality of movement is determined by the intrinsic mechanics of the arm+object, and will therefore not change significantly as a function of training, but that the central nervous system adapts its control strategy to make movements less sensitive to local perturbations. It is anticipated that the results of these experiments will lead to a better understanding of the neuromuscular control processes underlying object manipulation and will eventually help provide clearer direction for developing functionally realistic virtual interactions for motor rehabilitation.