Coordinated Natural Rhythmic Movements by Distributed Biological Oscillators

该项目的总体目标是了解动物在运动过程中观察到的自然节奏运动背后的生物控制机制，并为工程应用建立设计原则。特别是，我们重点研究了由神经元振荡器实现的分布式控制机制--中央模式产生器，旨在建立基于中央模式产生器的反馈控制器的设计理论，以稳健地实现机械系统的自然有节奏的运动。我们假设，对于闭环系统，通过在多自由度、轻微阻尼的机械系统的每个配置的传感器/执行器对之间的反馈回路中放置CPG单元，可以稳健地实现对自然振荡模式的卷吸。我们将利用多变量谐波平衡(MHB)的方法，建立控制设计参数的解析公式，以实现具有给定频率和振型的振荡。然后，我们将检验这一假设是否得到MHB分析预测的支持。反馈控制在许多工程应用中是必不可少的，在这些应用中，我们希望调整物理系统的运动。目前的技术可以在计算机控制的帮助下实现极其快速和准确的运动。然而，这种控制设备只擅长听从命令，缺乏在情况发生变化时“思考”该做什么的能力。例如，我们会在冰上以不同的方式行走，这样我们就不会滑倒或摔倒，但行走的机器人将无法调整其运动。这是因为运动是在假设名义情况下提前计划的，控制器迫使机器人按计划移动，即使情况发生变化。在生物系统中，CPG根据感觉反馈信息实时规划适当的运动。这项研究将揭示这种运动规划和控制集成背后的生物学机制，使设计出健壮和适应性强的工程系统成为可能。神经科学和控制工程之间的协同效应将加速这两个领域的发展。特别是，神经科学的知识将有助于揭示工程原理，而控制工程的结果将反过来为可以通过生理实验测试的生物机制提供新的预测。

The overall goal of this project is to understand the biological control mechanism underlying natural rhythmic movements observed during animal locomotion, and to establish a design principle for engineering applications. In particular, we focus on the distributed control mechanism realized by neuronal oscillators called the central pattern generators (CPGs), and aim to establish a design theory for CPG-based feedback controllers to robustly achieve natural rhythmic movements of mechanical systems. We hypothesize that entrainment to a natural mode of oscillation can be robustly achieved for the closed-loop system by placing a CPG unit in the feedback loop between each collocated sensor/actuator pair of a multi-degree-of-freedom, lightly damped, mechanical system. We will develop analytical formulas for the control design parameters to achieve an oscillation with a given frequency and mode shape, using the method of multivariable harmonic balance (MHB). We will then examine whether the hypothesis is supported by the predictions from the MHB analysis. Feedback controls are essential for many engineering applications in which we desire to adjust movements of a physical system. Current technologies enable extremely fast and accurate motions with the aid of computer-controls. However, such control devices are only good at following a command, and lack sophistication to "think" what to do when the situation changes. For instance, we would walk differently on ice so that we don't slip or fall, but a walking robot would not be able to adjust its motion. This is because the motion is planned ahead assuming a nominal situation, and a controller forces the robot to move as planned even when the situation changes. In biological systems, appropriate motions are planned in real time by CPGs, based on sensory feedback information. This research will uncover the biological mechanism underlying this integration of motion planning and controls, enabling designs of robust and adaptable engineering systems. The synergistic effect between neuroscience and control engineering will accelerate advance of both fields. In particular, the knowledge from neuroscience will help to uncover engineering principles, while the results from control engineering will in turn provide new predictions for biological mechanisms that can be tested through physiological experiments.