【作者】 黄海；

【导师】 刘宏；

【作者基本信息】 哈尔滨工业大学 ， 机械电子工程， 2008， 博士

【摘要】 传统的单自由度商用假手虽然也是一种典型的生物机电一体化系统,但只能进行简单的抓握,与人手还有很大的差距,远远满足不了残疾人正常生活和工作的需要。目前残疾人假手正朝着多自由度的方向发展,重量轻、体积小、可靠性高、自适应抓握,控制简单和操作灵活是假手的发展趋势。因此,本文结合国家自然科学基金重点项目“新一代仿人型残疾人假手系统及理论的研究”(项目编号:50435040),研制了达到国际领先水平的HIT-DLR新型仿人假手,并进行了单手指的运动学、静力学和动力学分析,动态轨迹跟踪及其柔顺控制的研究。HIT-DLR新型仿人假手,外形和人手相似,共有五个手指,十五个活动关节,通过三个步进电机驱动,总重量约为500g。根据欠驱动和耦合原理采用模块化思想设计了假手的手指,手指有很强的运动灵活性和可靠性。设计了由一个电机驱动的三指联动机构,该机构能完成对复杂物体的自适应抓取,同时该机构在外观和运动形式上实现了仿人化设计,具有较高的运动一致性和保持固有位姿的能力;设计了仿人拇指机构,该拇指在一个电机的驱动下,能够沿空间的锥面实现抓握,球轴承的使用使欠驱动原理在空间机构上得到了实现,同时为了保证抓握不同物体的有效性,用ADAMS进行了仿真实验,给出拇指的具体布置位置;设计基于应变测量的、可以互换使用的基关节力矩传感器。基于集成化的思想设计假手的本体结构,实现假手的机构、传感、驱动和微处理器系统的集成,完成了假手的外包装设计,包括假手外观美化设计和假手的机构优化设计。假手的抓取试验和负载实验验证了假手的设计思想。进行了食指连杆机构和拇指空间连杆机构的运动学分析,完成手指耦合四连杆的参数设计,确定连杆的机构参数,并在ADAMS虚拟环境下进行了仿真验证。建立假手欠驱动单手指的静力学模型,给出抓握不同物体时驱动力矩和各指节受力之间的关系,使用ADAMS进行了验证,并使用压力传感器进行了实验验证。基于手指,由MATLAB和ADAMS进行了仿真、比较,得到了一致的结果,并进行了实验验证。假手的手指控制性能直接影响假手的操作性能,是假手设计的重要一环。单手指可以看作一个小机器人,手指的运动学和动力学分析,也是为手指的控制做准备。在控制方面,首先结合假手的控制系统平台,进行了计算力矩法的轨迹跟踪控制实验,其结果大大减小了PID轨迹跟踪的误差,提高了动态性能,取得了良好的轨迹跟踪效果,消除了在运动控制中扭簧造成的不可控性;由于假手手指只有位置传感器,这就不可避免地给控制带来了的误差,为了减少这种误差,建立了基于假手手指动力学模型的速度观测器,该观测器运用在手指的控制中可以通过手指的位置传感器信号和动力学模型准确的获得速度信号,弥补了假手没有速度传感器的缺点,大大减小了控制误差。加入速度观测器的计算力矩轨迹跟踪算法,进一步减小了跟踪误差;结合了速度观测器的自适应控制器,补偿动力学模型的中的不确定因素,减小了速度误差,使手指的动态控制效果比较理想。当假手在完成与环境接触的作业时,手指的柔顺性十分重要,阻抗控制是实现手指主动柔顺的主要方法之一,得到深入研究和广泛应用。本文分别通过基于位置和基于力的阻抗控制对手指基关节的抓握力进行控制,其中在基于力的阻抗控制中同样也使用了速度观测器。基于力的阻抗控制通过动力学补偿不仅可以准确地进行力跟踪,而且把轨迹跟踪和力跟踪结合起来,实现了手指的动态抓握控制。更多还原

【Abstract】 Although traditional commercial 1-DOF prosthetic hand is a typical biomechatronics system, it can only complete simple grasp which is far different from human hand and can’t satisfy handicapped normal life. Nowadays, the general tendency of prosthetic hand is multi-DOF, light-weight, small-volume, high-reliability, easy control methods and satisfactory manipulative performance. Therefore, based on the key program of National Natural Science Foundation of China (NSFC)“Study on humanoid prosthetic hand system and its theory”(No. 50435040), HIT-DLR Prosthetic Hand II on international top is developed; kinematics, statics and dynamics of index finger are analyzed; dynamic based curve fitting and impedance based force control are researched.The biomechatronics HIT-DLR Prosthetic Hand II possesses similar externality of human hand, five fingers and 15 active joints. It’s actuated by 3 step-motors and weight 500g. Based on under-actuated and coupling principle, the fingers are designed with high agility, reliability and modularization idea. The three finger transmission scheme is developed. Actuated by one motor, the scheme can make the mid finger, ring finger and little finger complete auto-adapted grasp for complex shaped objects. It can grasp coherently and stay original posture. The bio-thumb is designed. It can grasp along a cone surface actuated by a motor. The principle of under-actuate is realized in spatial linkages mechanism through using ball bearing. ADAMS simulations are performed for thumb position in order to guarantee its successful grasp for different objects. The calibrated torque sensor with stress-measuring which can interconvert is designed to measure base joint torque. Structure of prosthetic hand is designed with modularization and integration thinking, and the integration of structure, sensor, controlling and driving circuit system of the prosthetic hand is realized. The envelop designation is accomplished. It includes external perfection and mechanism modification. The grasp and loading experiments verify the designation ideas.Kinematics of linkages of index finger and spatial linkages of thumb are analyzed. The parameters design of coupling linkages are completed, decided and simulated in ADAMS. The statics model of index finger is constructed in order to determine the relationship between actuation torque and the support force from phalanges, which is verified through ADAMS simulation and experiments. Based on virtual spring approach, the dynamic analysis of under-actuated finger is achieved. The index finger is dynamically modeled in this way. Coherent results are obtained through MATLAB and ADAMS simulation. Experiments are performed to verify the dynamic analysis.As one of key issues of prosthetic hand, the performance of finger control plays an important role in the hand manipulation. A finger may be treated as a small robot, and the theories about robot kinematics and dynamics are also prepared for the finger control. Based on the platform of prosthetic hand control system, Experiments of curve fitting using computed torque have been performed. The experiments has eliminated the uncontrollable character in PID control, reduced fitting errors greatly, improved its dynamic performance and achieved good results. On the other hand, control with only signals of position sensors must brings out errors. To reduce control errors, a velocity observer based on dynamic model of the finger has been established and used in improving computed torque curve fitting algorithm. To complement the uncertain factors of dynamic model, the adaptive controller with velocity observer has been designed. Velocity errors have been reduced and ideal results of dynamic control have been achieved.When prosthetic hand works, the finger compliance is very important. As one of main ways realizing compliance, the impedance control is studied deeply and implemented widely. Position and force based impedance control have been researched through the base joint sensors. Therefore, the grasp force of finger’s phalanges can be controlled through its base joint torque control. The velocity observer has been used in impedance control. Compensated with inverse dynamic equation, the force based impedance control can not only realize accurate force tracking, but achieve finger’s dynamic control by the combination of curve fitting and force tracking.更多还原