【作者】 李正义；

【导师】 唐小琦；

【作者基本信息】 华中科技大学 ， 机械电子工程， 2011， 博士

【摘要】 随着机器人应用领域的不断扩展和对机器人智能化要求的不断提高,以往以位置控制为主的机器人控制方法已不能满足某些复杂环境(装配、抛光、去毛刺、助力机器人、康复机器人)的应用要求,机器人与环境间力/位置控制方法同时实现机器人末端接触力和位置期望值跟踪控制,适应了机器人在复杂环境中运动控制的要求；当前,机器人与环境间力/位置控制研究大都基于机器人阻抗控制和机器人力/位置混合控制。本文在国家科技重大专项和国家自然科学基金课题的资助下,对机器人与环境间力/位置控制关键技术开展研究,主要内容和创新点有：常规机器人阻抗控制性能受环境动力学和期望阻抗模型参数的影响明显,为实现不同环境下机器人阻抗控制性能保持不变,期望阻抗模型必须作相应调整：在定义机器人阻抗控制性能指标基础上提出系统稳态时阻抗模型刚度计算方法；结合神经网络估计环境等效刚度和二阶系统临界阻尼条件,计算出阻抗模型阻尼初值；提出了自适应模糊控制方法根据测量的接触力、位移误差及其导数实时调整阻抗模型的阻尼、刚度参数值,提高机器人阻抗控制中接触力/位置控制的动静态性能；此外,为保持机器人与环境间接触状态和实现期望的接触力控制,给出了环境位置误差允许的范围；提出了机器人操作空间中基于伪速度的机器人位置控制方法,提高机器人运动控制动态性能。修改常规的机器人力/位置混合控制方法,提出了机器人位置、姿态、力和力矩混合控制方法；针对机器人力/位置混合控制易受系统干扰影响的问题,研究基于Kalman状态观测器的机器人力控制方法。结合机器人位置、姿态、力和力矩混合控制和Kalman状态观测器的优点,提出了性能优良且对系统干扰具有较强抑制能力的机器人力/位置混合控制方法。将机器人与大刚性环境间碰撞接触过程依次分为：接近运动阶段、冲击振荡阶段、阻尼振荡阶段和稳定阶段,为实现机器人末端与环境间快速稳定和期望的接触力控制,设计了针对不同阶段的控制策略：针对接近运动阶段的自适应非接触阻抗控制,通过神经网络实时调整阻抗模型参数以改变机器人接近过程的动力学特性,在实现快速接近前提下减小后续的碰撞接触冲击效应；针对冲击振荡阶段的机器人位移、速度与驱动力关系,设计镇定控制器抑制机器人末端位移回弹运动以减小冲击振荡幅值,缩短冲击振荡持续时间；针对阻尼振荡阶段,提出了变参数机器人力/速度控制方法,建立基于系统能量快速衰减的参数调整规则,加快阻尼振荡的稳定和实现期望的接触力控制。综合冲击振荡与阻尼振荡阶段控制方法,设计模糊控制器根据镇定控制原理对机器人力/速度控制中参数进一步微调,实现碰撞接触后系统的快速稳定和期望的接触力跟踪控制。抛光机器人控制系统设计,提出了基于机器人力/位置混合控制和阻抗控制原理的机器人抛光作业中抛光力和位置控制方法：采用阻抗模型跟踪控制实现接触面法向方向抛光力跟踪控制；对于接触面切向方向的进给速度控制,采用模糊控制实现基于刀具轨迹的接触而切向进给速度前馈控制；此外,设计了基于现场总线的抛光机器人控制系统软硬件结构,机器人抛光实验验证控制方法的有效性。更多还原

【Abstract】 With the continuous expansion of the application area of robotics and the sustained growth of the robot intelligent level request, the position tracking control pointed robot control system can no longer meet the application in complex environments(assembly, polishing, deburring, human augmentations, rehabilitation robotics). The position force control method for the robot-environment interaction is aimed at establishing robotic end-effector force and position tracking control simultaneously, it meets the requirements of robot motion control in the complex environment. At present,studies on force position control between robot and environment mainly focus on the robot impedance control and the hybird position/force control of a robot.This thesis, supplied by the National Natural Science Foundation, the National Science and Technology major project, specializes the key technologies of the robot force position control method for the robot-environment interaction, the main content and innovation of the thesis is as follows:The performance of conventional robot impedance control mainly depends upon the environment dynamics and the choice of the expected impedance model. To maintain the performance of the robot impedance control in a wide range of environments, the expected impedance model needs to be adjusted adaptively. The caculation of the spring parameter of the impedance model in system steady-state conditions is based on the self-defined force impedance control performance index. Combined with the environment spring equivalence value estimated by neural networks and the second-order systems critical damping condition, the damping initial value of the impedance model is determined. In order to improve the force position tracking static-dynamic performance, the self-turning fuzzy control method is designed to achieve adjustment for the damping and spring parameters of the impedance model based on the measured force, position and its derivative in real time. Moreover, the environment position tolerance for maintaining contacted status and establishing expected force tracking control for the robot-environment interaction is illustrated.The quasi-velocity-based robot position control method for a robot in the operational space is also proposed to improve the dynamic performance of the robot motion control.The hybrid position, posture, force, and moment control method of a robot is proposed by the modification of the conventional hybrid position/force control. As for the question of the hybrid position/force control of a robot susceptible to system disturbance, the robot force control method based on Kalman active observers is proposed. Moreover, combined with the advantages of Kalman active observers and the hybrid position, posture, force and moment control method, the designed hybird position/force control of a robot exhibits good performance and is robust to system disturbances.The robot’s physical impact and contact with the stiff environment is divided into approaching movement, impact oscillation, damping oscillation and steady stage successively. In order to stabilize the robot’s physical impact and contact with an environment and achieve the expected contact force control, control strategies are designed for different phases:the robot noncontact impedance control method is designed for the approaching movement and the impedance parameters are adjusted by neuron networks for changing robot dynamics characteristics during the approaching movement, it decreases the succeeding impact/contact effect on the premise of fast approaching;in order to diminish the amplitude of the oscillation and shorten the impact duration time, the designed suppression controller represses the position rebound based on the robot end-effector’s displacement, velocity and driving force in the impact oscillation phase, the varying-parameter robot force/velocity control method is designed to improve the stability of the damping oscillation phase based on the fast rate of energy dissipation of the system. Meanwhile, combined the control methods for the impact oscillation and damping oscillation, a fuzzy control method is designed to facilitate fine adjustments of the system force feedback coefficient based on the suppression control principle, it improves the force tracking control performance and stability of the impact/contact transition processes for the interaction between the robot and environment in contact.In the design of the polishing robot control system, the polishing force and position control is implemented by integrating the hybird position/force control and the robot impedance control principle:the designed impedance model following control method is adopted to realize the polishing force control in the constrained surface normal direction, the fuzzy logic controller plans the speed feed-forward in the constrained surface tangential direction based on the tool path data. In addtion, the realization of the software and hardware structure of the polishing robot control system based on the fieldbus is introduced.The feasibility and effectiveness of the control methods is confirmed by some experiment.更多还原