【作者】 冯治国；

【导师】 钱晋武；

【作者基本信息】 上海大学 ， 机械电子工程， 2009， 博士

【摘要】 助行腿机器人系统是外骨骼机器人技术与减重康复训练相结合的产物。它利用外骨骼机器人动作精度高、响应速度快、“不知疲倦”的特点,有可能彻底解决人工手动减重训练强度难以保证、精度不够、训练数据难以反馈等问题,并将理疗师从繁重的体力劳动中解放出来,使其可以将更多的精力专注于患者康复训练效果的评估和康复计划的制定上,从而提升患者步行康复训练的质量和效率。目前,步行训练外骨骼机器人已成为国外神经康复技术的重要发展方向之一。本课题是上海大学机电工程与自动化学院智能机械与系统研究室承担国家863项目“步行训练机器人系统关键技术”的子课题,论文对外骨骼助行腿机器人系统的机构设计、系统建模、步行运动训练控制策略和实验验证等关键技术展开研究,具体研究内容如下:⑴基于满足步行训练的功能要求以及安全性的考虑,结合人体工程学、仿生学和机械设计等技术,设计具有髋、膝和踝关节的3自由度的连杆助行腿。采用自行设计的电动直线驱动器动力装置,实现外骨骼助行腿髋、膝关节主动屈伸运动以及踝关节的主动跖屈和背屈运动,并制作了实物样机试验系统。⑵在机器人辅助训练模式与康复治疗对应的原则下,设计“机器主动”和“患者主动”运动训练控制模式。采用拉格朗日法,建立机器主动训练模式下人机系统的动力学模型。为研究训练者与助行腿之间的人机耦合运动,利用牛顿–欧拉法构建患者主动训练模式的人机系统动力学方程。⑶针对“机器主动”运动训练控制模式,采用计算力矩加比例–微分反馈控制算法,分析了建模误差以及外界干扰等不确定性因素对计算力矩加比例–微分反馈控制算法的影响,推导证明算法的收敛性。为消除建模误差影响,提高系统步态轨迹的跟踪能力,引入径向神经网络补偿建模误差。对“患者主动”运动训练控制模式,设计了基于位置阻抗控制方法,从理论上对建模误差与阻抗关系进行了分析。⑷通过研究了SolidWorks、ADAMS和MATLAB/Simulink三种软件集成的协同仿真方法,建立了助行腿机器人虚拟样机协同仿真平台。在平台上,进行助行腿运动学和动力学仿真分析,步态轨迹跟踪控制算法仿真实验。仿真结果验证了运动学和动力学模型理论分析,为助行腿机构优化和驱动器的电机选型提供重要的参考依据。仿真实验数据表明计算力矩加比例–微分反馈控制算法对助行腿轨迹跟踪控制是有效的,为该算法在实际样机的应用奠定基础。⑸在助行腿机器人样机系统实验平台上,进行了一系列的功能和性能实验,测试了助行腿机器人系统性能的稳定性、安全性和可靠性,并对试验结果进行分析。指出了存在的不足,为进行受试者参与系统测试实验做了一定准备。本论文在以下方面进行了新的探索并取得成果:⑴探索性研究踝关节主动康复训练,导致助行腿的结构复杂化和加大助行腿运动控制的难度,通过利用多轴运动控制系统可以解决这个控制难度问题。带有踝关节主动康复训练的助行腿机器人可使下肢运动康复训练更加全面,符合临床习惯要求。⑵建立了“机器主动”运动训练控制模式下助行腿机器人系统在跑步机上步行的动力学模型,采用主从跟随控制思想,设计计算力矩加比例–微分反馈的控制算法和径向基函数的神经网络补偿控制的方法,弥补机器人动力学模型的不确定性,提高助行腿轨迹跟踪能力。⑶通过将训练者与助行腿隔离分析研究,采用牛顿–欧拉法分别建立动力学模型,利用人机交互作用信息建立训练者与助行腿组成的耦合系统的解耦关系,为研究“患者主动”训练模式的控制方法奠定理论基础。本博士论文深入研究步行训练机器人系统关键问题之一——助行腿。通过对助行腿关键技术的研究,为发展面向应用的步行训练机器人系统提供必要的理论依据、实验数据和研究经验。随着相关技术不断发展完善,将步行训练机器人技术转化为机器人产品,这将对于提高神经受损患者的康复效果和质量、具有积极的学术意义和重要的实际意义。更多还原

【Abstract】 An exoskeleton robot for motor training in gait rehabilitation is in combination with the robot technology and the body weight supported treadmill training. When introduced into the gait rehabilitation after spinal cord injury, the exoskeleton robot performs with the characteristics of high accuracy, fast response and "tireless", and it is likely to avoid the drawbacks of the traditional body weight supported treadmill training. Meanwhile, the exoskeleton robot will help to train the patients as well as record the training data, free the doctors from the heavy physical work so that they can have more time to focus on the superior work such as rehabilitation training evaluation and making recovery plans. And then, the quality and efficiency of the rehabilitation training will be highly enhanced. Thus far, the exoskeleton robot has been a focus on the development of neuro-rehabilitation technique in the worldwide.The research subject on key technologies of the rehabilitation robot in gait training, which is sponsored by National High-Tech R & D Program, is undertaken by the Laboratory of Intelligent Machine and System at School of Mechatronics Engineering and Automation of Shanghai University. Some research works associated with the exoskeleton robot are described in this thesis, including mechanical design, system modeling, method of motion control, etc. They are described in detail as follows:Considering the patients’requirements and safety of gait training, the exoskeleton robot is designed combination of many technologies, such as ergonomics, bionics and mechanical design. Each leg of the exoskeleton robot has three degree of freedom at hip, knee and ankle joints. To drive flexion and extension movement of joints, a custom-designed electric linear actuator is adopted. A physical prototype of exoskeleton robot has been developed. Robot-in-charge and patient-in-charge modes are introduced on the principle of rehabilitation therapy and robot-assisted training. Using the Lagrange method, the mathematic model in robot-in-charge mode is given. In order to develop man-machine coupling between the trainer and the exoskeleton robot, dynamics equations of man-machine system in patient-in-charge mode is built based on the Newton-Euler approach.The computed torque control law with proportion-differential feedback is designed in robot-in-charge mode. Uncertain factors of dynamic model influencing the control algorithm are discussed and the convergence of the control method is proved in theory. To remove modeling error and improve trajectory tracking control effect, a compensating scheme on radial-basis-function neural network is presented. In patient-in-charge mode, a position-based impedance control approach is introduced. The relation between modeling error and impedance is analyzed theoretically.To develop the exoskeleton robot, a method based on virtual prototype and collaborative simulation is proposed. Solidworks, Adams and Matlab/Simulink are integrated to be used to establish united simulation platform for the exoskeleton robot. Kinematics and dynamics simulation of the exoskeleton robot, the trajectory tracking control laws are done in collaborative simulation platform. The results give important reference for mechanism optimization and motor selection, indicating that the proposed method can control and track the trajectory of the exoskeleton robot effectively.Through establishing a prototype of the exoskeleton robot, a series of training experiments are carried out, which demonstrate the functionality, safety and reliability of the exoskeleton robot. The results prove the feasibility of the exoskeleton robot system. But the exoskeleton robot still need be improved for human testing.The achievements of this thesis are summarized as follows:An actuated robot ankle joint applying to rehabilitation training may result in the complex structure of the exoskeleton robot and the difficulty of motion control which can be solved using multi-axis motion control system. The exoskeleton robot with the actuated ankle joint makes patients trained completely, in accord with the requirements of clinical practice.The dynamics model of the man-machine system walking on the treadmill in robot-in-charge mode is built. Based on the master-slave following control strategy, a computer torque control law with proportion-differential feedback and a compensating scheme based on radial basis function neural networks are designed to make up uncertain factors of the dynamic model and increase the ability of the trajectory tracking.Dynamics equations of the trainer and the exoskeleton robot are built respectively based on the Newton Euler approach. The decoupling relationship between the trainer and exoskeleton robot is found using man-machine Interaction Information. This provides theoretical foundation for developing dynamic control methods in patient-in-charge mode.The exoskeleton robot, one of key issues on rehabilitation robot in gait training, is developed deeply in this thesis. These works provide the necessary theoretical basis, experimental data and valuable research experience for development of rehabilitation robot in gait training. With improvement of the related technologies, some products on the rehabilitation robot are realized. This is of positive academic significance and practical importance to improve the quality of rehabilitation training.更多还原