Dimensionality Reduction in the Control of the Human Hand

这个项目的目标是描述手部运动控制中的降维，并确定降维背后的神经控制原则。采取综合的方法，结合非侵入性的人体实验与控制和信息理论的方法。理论方法是对实验的补充，也是对神经活动和手部行为不同层次描述的桥梁。研究计划由两部分组成。第一部分是识别运动基元，即手部运动的基本构建块。这一部分的一个重要组成部分是建模的层次神经网络负责形成和实现的运动原语。第二部分采用信息论的方法，得到在外围和中心约束下手控制的灵活性和信息容量的指示。这两个部分提供了同一问题的不同观点，但互补的工作，以促进对手部神经控制降维的理解。这项研究的动机是人手的神经控制。手具有大量的机械自由度，这为手执行熟练的手指运动提供了巨大的灵活性。然而，另一方面，手的灵活性使得控制问题非常具有挑战性。神经系统如何处理手部控制的高维性仍然是一个未解之谜。该研究旨在促进对手部运动控制的神经组织和降维机制的理解。在这个方向的研究将是有价值的灵巧机器人，类脑电路和脑机接口的设计与各种潜在的应用在工业，国防和医学，例如，在开发假肢，可以帮助中风患者，残疾人，和那些与大脑或脊髓损伤。此外，受处理高维问题的神经原理的启发，本研究将为大规模工程系统的控制提供见解。

The goal of this project is to characterize the dimensionality reduction in the control of hand movements and to identify the neural control principles underlying the dimensionality reduction. A comprehensive method is taken that combines non-invasive human experiments with control and information theoretic approaches. The theoretic approaches complement the experiments and bridge different levels of description of neural activities and hand behaviors. The research plan consists of two parts. The first part is to identify the movement primitives, i.e. the fundamental building blocks, of hand movements. An important component of this part is to model the hierarchical neural networks responsible for the formation and implementation of movement primitives. The second part takes an information theoretic method to obtain indication for the flexibility and information capacity of the hand control under peripheral and central constraints. These two parts offer different views of the same problem, but work complementarily to promote the understanding of dimensionality reduction in neural control of the hand. The research is motivated by the neural control of the human hand. The hand has a large number of mechanical degrees-of-freedom, which offers tremendous flexibility for the hand to perform skilled finger movements. On the other hand, however, the flexibility of the hand makes the control problem very challenging. It is still an unsolved mystery how the neural system handle the high dimensionality underlying the hand control. The proposed research aims to promote the understanding of neural organization and mechanism for the dimensionality reduction in the control of hand movements. Study in this direction will be valuable for the design of dexterous robots, brain-like circuits, and brain-machine interfaces with a variety of potential applications in industry, defense, and medicine, for example, in developing prosthetics that may aid stroke victims, handicapped individuals, and those with brain or spinal cord damages. Furthermore, inspired by the neural principles for handling problems of high dimensionality, this research will provide insights for the control of large-scale engineering systems.