基于神经网络的水下灵巧手阻抗控制

水下灵巧手抓取物体时,物体与指尖存在力控制问题,但是由于动力学模型、被抓取物体位置和刚度的不确定性,采用传统阻抗控制方法不具有鲁棒性.文中基于位置型阻抗控制方法,提出采用神经网络对手指动力学模型、物体刚度和物体位置误差进行补偿.对补偿策略进行了详细的推导,并通过仿真实验验证了该方法的补偿效果,结果表明基于神经网络的位置型阻抗控制器具有较强的鲁棒性.     

DLR/HIT多指机器人灵巧手的研究

基于机电一体化设计理念和集成化、模块化设计方法研制了DLR/HIT灵巧手,DLR/HIT灵巧手具有4个相同结构的模块化手指,共具有13个自由度,具有拟人手形的外观.采用商品化的直流无刷电机实现了手指的驱动;手指具有位置、力/力矩及温度等多种感知能力;基于DSP/FPGA等实现了手指的通讯和实时控制;高度集成,将所有部件均集成在手指和手掌内.DLR/HIT手具有很好的可靠性、可操作性和精细操作能力.     

基于机电一体化的机器人灵巧手手指的研制

运用机电一体化设计思想,设计了新一代灵巧手手指,具有3个自由度,由3个体积小、出力大的无刷直流电机驱动,机械传动系统采用刚性的减速器、齿轮及连杆传动,使手指具有很好的刚度.该手指具有丰富的感觉功能,包括位置感觉、力/力矩感觉以及温度感觉等.该手指集机械本体、驱动、传感器及电路为一体,最大限度地实现了灵巧手手指的集成化、模块化.     

模块化嵌入式五指机器人灵巧手手指控制系统

为开发高度集成的五指机器人灵巧手,提出了采用DSP(Digital Signal Processor)+FPGA(Field Programmable Gate Array)控制结构的模块化嵌入式手指控制系统设计方法。实现了采用单电流传感器检测全桥控制下相电流及相电流过流双保护。为解决手指DSP和FPGA之间如何稳定、高效通信的难题,结合先进先出FIFO(First In First Out)寄存器,设计了基于串行通信接口SCI(Serial Communications Interface)和RS485总线的多中断差分通信系统。实验证明,这种DSP+FPGA控制结构及模块化的设计,使得手指控制部分得以高度集成,同时DLR/HITII灵巧手获得很好的整体性能和稳定性。     

基于阻抗模型的卧式康复机器人柔顺控制

为满足脑卒中患者早期卧床康复需求,结合临床康复手法与人机工程学,设计了一款卧式康复机器人。该卧式康复机器人可以提供单腿屈髋屈膝运动、卧式步态运动、桥式运动等多种有效的康复动作。为确保训练的安全性以及改善在训练过程中由人机交互作用带来的舒适度和柔顺性差等问题,采用一种基于阻抗模型的柔顺控制方法。将人与机器人之间的交互视为一种虚拟的阻抗模型,即二阶质量-弹簧-阻尼模型,并基于该阻抗模型进行康复机器人的柔顺控制研究。该柔顺控制策略由外环阻抗控制器和内环PID控制器组成,外环的阻抗控制器通过将人机交互力作用在阻抗模型上,实现根据人体意图对运动轨迹进行改变,而内环的PID控制器主要实现对生成的期望轨迹进行稳定跟踪。通过实验证明了基于阻抗模型的卧式康复机器人柔顺控制的有效性。     

漂浮基空间机器人姿态与末端爪手协调运动的反演滑模控制

讨论了载体位置无控、姿态受控情况下,具有外部扰动的漂浮基空间机器人系统的控制问题。结合系统动量守恒关系及拉格朗日方法,建立了漂浮基空间机器人的运动Jacobi关系及系统动力学方程。以此为基础,借助于反演设计方法,设计出空间机器人载体姿态与机械臂末端爪手协调运动的滑模控制方案。此方案可有效地减少控制系统的在线计算时间,更能满足空间机器人这一时变系统控制的实时、在线要求。数值仿真结果证实了该控制方案的有效性。     

基于自适应虚拟阻抗的微电网无功均分控制

在微电网多逆变器并联系统中,由于微电网各逆变器之间存在输出阻抗和馈线阻抗不匹配现象,采用传统的下垂控制方法难以根据下垂系数合理地进行无功分配,同时还会产生无功环流问题。为提高无功均衡的精度,提出了一种基于自适应虚拟阻抗控制的控制策略。该方法在固定虚拟阻抗的基础上增加了积分环节,结合了改进的无功功率需求值来自适应调节分布式电源单元,最终补偿馈线阻抗失配,从而精确均分无功功率。此控制策略不仅无需提前预知馈线阻抗值,而且不需要通讯平台的参与,可在结构复杂的微电网中应用。研究结果表明:提出的虚拟阻抗无功均分控制方法相比于传统虚拟阻抗方法,其自适应能力和无功均衡性能更佳。     

多关节机器人的自适应阻抗控制研究

为了对动车组侧窗玻璃安装机器人末端接触力进行控制研究,在阻抗控制的基础上,提出了一种多关节机器人自适应阻抗控制算法,该算法能实现机器人末端接触力准确跟踪期望力。以PUMA560机器人前三关节为对象在接触空间进行仿真研究,仿真结果表明,基于自适应阻抗控制方法能很好地对动车组侧窗玻璃安装机器人的位置和力进行跟踪。     

Cartesian impedance control of dexterous robot hand

Presents a novel compliant motion control for a robot hand using the Cartesian impedance approach based on fingertip force measurements. The fingertip can accurately track desired motion in free space and appear as mechanical impedance in constrained space. In the position based impedance control strategy, any switching mode in contact transition phase is not needed. The impedance parameters can be adjusted in a certain range according to various tasks. In this paper, the analysis of the finger‘s kinematics and dynamics is given. Experimental results have shown the effectiveness of this control strategy.     

On set-point tracking control of dexterous robot hand

A model-free set-point tracking control approach of multi-fingered robot hand is presented.The set-point tracking controller,which has the structural form of PD controller,is composed with a combination of feedforward term,feedback term and saturation control term.The controller does not require the explicit use of dynamic modeling parameters.Experiments performed on the HIT/DLR hand demonstrate the effectiveness of the proposed approach in performance improvement and real-time application.     

Design and control of integrated pneumatic dexterous robot finger

Based on flexible pneumatic actuator(FPA),bending joint and side-sway joint,a new kind of pneumatic dexterous robot finger was developed.The finger is equipped with one five-component force sensor and four contactless magnetic rotary encoders.Mechanical parts and FPAs are integrated,which reduces the overall size of the finger.Driven by FPA directly,the joint output torque is more accurate and the friction and vibration can be effectively reduced.An improved adaptive genetic algorithm(IAGA) was adopted to s...     

Design and control of integrated pneumatic dexterous robot finger

Based on flexible pneumatic actuator (FPA), bending joint and side-sway joint, a new kind of pneumatic dexterous robot finger was developed. The finger is equipped with one five-component force sensor and four contactless magnetic rotary encoders. Mechanical parts and FPAs are integrated, which reduces the overall size of the finger. Driven by FPA directly, the joint output torque is more accurate and the friction and vibration can be effectively reduced. An improved adaptive genetic algorithm (IAGA) was adopted to solve the inverse kinematics problem of the redundant finger. The statics of the finger was analyzed and the relation between fingertip force and joint torque was built. Finally, the finger force/position control principle was introduced. Tracking experiments of fingertip force/position were carried out. The experimental results show that the fingertip position tracking error is within ±1 mm and the fingertip force tracking error is within ±0.4 N. It is also concluded from the theoretical and experimental results that the finger can be controlled and it has a good application prospect.     

一种教学机器人控制系统研发

为满足开放式教学的要求,设计了一种教学机器人,提出了控制系统的总体结构,设计了控制系统的硬件和软件,实验表明此控制系统能够较好地控制教学机器人．     

基于虚拟弹簧的欠驱动手指的动力学建模和控制

采用虚拟弹簧对2关节欠驱动手指进行了动力学建模。并使用该方法建立实际的3关节欠驱动手指的动力学模型,通过动态控制实验验证了动力学模型的正确性。该方法不仅避免了求解微分几何方程,并且直接派生出可解耦的动力学模型。可直接进行逆动力学分析、仿真和实时控制。同时,建立了基于动力学模型的速度观测器,用于轨迹跟踪,弥补了欠驱动手没有速度传感器的缺点,补偿了欠驱动环节造成的不确定因素。与PID或计算力矩法相比较,其轨迹跟踪误差更小,动态控制效果更好。     

Design and modeling for a kind of humanoid dexterous hand with elastic palm

With dexterous hands, robots can improve the work scope and work ability significantly. As palms of the existing multi-hand robots are made of steel plates that have small contact area, the robots cannot grab firmly. In this study, a new five-fingered dexterous robot hand is developed. Having flexible palm with 17 degree of freedoms （ DOFs）, the hand can grasp more stably and firm- ly. First, the forward kinematics and inverse kinematics of the fingers and the hand are calculated. Then, the connection between the force exerting on the end effectors and the torque exerting on the joint is set up, laying the foundation for the following control. Finally, through the analysis and sim- ulation of the position, velocity and acceleration, the trajectory planning has a better performance.     

Fault tolerant control for a robot collaborative system

A new fault-tolerant control scheme is proposed for a nonlinear collaborative system that contains two robot subsystems. When fault occurs in one subsystem, the fault-free subsystem is used to compensate the fault influence of the faulty one on the whole collaborative system. When the faulty subsystem could not repair itself or the repair process needs a long time, the controller of the fault-free subsystem is reconfigured using the fault diagnosis information and other measured information, leading to the fault tolerant control of the robot collaborative system. Simulations of fault tolerant control for the robot collaborative system show the effectiveness of the proposed method.