【作者】 江侃；

【导师】 李炜；

【作者基本信息】 华中科技大学 ， 控制工程， 2013， 硕士

【摘要】 运动是人体与外界环境交互作用的唯一途径，也是人体的一项基本功能。经过不断地进化与完善，人体运动系统是已知的最智能的系统。在处理运动过程中存在的不确定性、高噪声等问题时，运动控制系统表现出良好的鲁棒性及自适应性，能够通过运动学习而逐渐适应外部环境，本文的目的就在于研究其学习及自适应机制。本文首先介绍了运动控制研究的相关背景知识、研究的目的及意义，同时综述了该领域的国内外研究概况。在前文论述的基础上，提出了运动控制系统对于重复干扰具有适应性，对随机干扰无适应性以及对重复干扰表现出的自适应性是基于认知调控的假设，以该假设为基础进行了一个带有干扰的到达-抓握实验。通过分析记录到的皮层神经元活动以及运动轨迹，发现运动系统对随机干扰没有表现出适应性，对于重复干扰表现出明显的学习及适应过程，最终能够形成一种预测控制策略，即实验结果证明了假设的正确性。随后，提出一个运动控制的假设模型来解释该实验结果，整个模型由运动规划及运动执行组成。其中模型的运动规划部分基于贝叶斯决策框架，运动规划的结果是系统以不同的权重调用存储在运动记忆中的固有模式，权重由贝叶斯决策的结果确定，整个调用过程可以用一个线性回归方程来表示。最后，通过记录到的运动轨迹求解回归方程参数，计算结果与假设模型预测结果一致，证明了假设模型的合理性。更多还原

【Abstract】 Motor is the only way to interact with the external environment as well as the basicfunction for human body. At present, human motor control system that experience a longevolution process is the most impeccable and intelligent control system. When faced withthe uncertainty and high noise exist in the motor system，it displays favorable robust andadaptive and it can gradually adapt to the external environment by motor learning, thepurpose of this thesis is to research on the mechanisms of motor learning and adaptation inmotor control system.In this thesis, we first introduce the background，purpose and meaning of motorcontrol and review the study situation in and out of China of this field. Based on this, weproposed a hypothesis that the motor control system can adapt to repeat perturbations andcan’t adapt to random perturbations and moreover, the adaption to repeat perturbations isbased on the modulation of cognition. Based on the hypothesis, we performed areach-to-grasp experiment with perturbations. By analyzing the activity of motor corticalneurons and movement trajectories recorded in the experiment, we find the motor systemdisplays non adaption to random perturbations and an obvious learning and adaptionprocess and develop a predictive control strategy at last, i.e., the experiment results provethe correctness of the hypothesis.We proposed a hypothetical motor control model composed of motor plan and motorexecution to explain the results. The motor plan in this model is based on Bayesiandecision theory framework and the result of it is that the system will invoke the inherentpatterns stored in motor memory with different weights, which are determined by theresults of Bayesian decision. A linear regress equation was used to represent the wholeinvoking process. At last, the movement trajectories were used to calculate the parametersof this regress equation, the calculations are consistent with the preconceived results of hypothetical model, which can prove that the hypothetical model is reasonable.更多还原