全局和声搜索方法及其在仿人灵巧臂逆运动学求解中的应用

仿人灵巧臂逆运动学(IK)问题可转化为等效的最小化问题,并采用数值优化方法求解.和声搜索(HS)是模拟乐师在音乐演奏中调整音调现象的一种启发式搜索方法,目前还尚未在机器人机械臂逆运动学问题中得到应用.本文提出一种基于粒子群体智能的全局和声搜索方法(GHSA),该方法在和声搜索算法中引入微粒群操作(PSO),采用粒子群策略替代常规和声搜索算法中的搜索法则创作新和声,通过粒子自身认知和群体知识更新和声变量位置信息平衡算法对解空间全局探索与局部开发间能力;同时算法还引入变异操作增强算法跳出局部最优解能力,基准函数测试表明该方法改善了全局搜索能力及求解可靠性.在此基础上以七自由度(7-DOF)冗余仿人灵巧臂为例,考虑以灵巧臂末端位姿误差和“舒适度”指标构建适应度函数并采用GHSA算法求解其逆运动学(IK)问题,数值仿真结果表明了该方法是解决仿人灵巧臂逆运动学问题的一种有效方法.     

具有自适应弹射机制的粒子群算法

针对粒子群算法容易陷入局部最优和停滞的问题,提出自适应弹射机制的粒子群算法.为了保持粒子群的活力,在算法内引入弹射操作.当粒子满足条件,当前位置赋予很大的速度,使其飞到很远的区域.弹射方式可以选择全维弹射和概率弹射.为了配合弹射操作,提出粒子优劣的判断机制,使粒子可以被弹射飞出可行域.在算法中设定自适应判别函数,当粒子满足该判别函数,对粒子实施弹射.数值实验表明,文中算法具有较强的全局搜索能力和较快的搜索速度.     

基于精英策略粒子群算法机械臂逆运动求解

多关节机械臂空间姿态控制研究中,逆运动分析是控制的基础,由于多关节机械臂逆运动求解问题较为复杂,对其求解结果优劣性的判别比较困难.为了解决传统关节变量算法在求解过程中收敛精度不高、局部最优解过多的问题,利用精英策略搜索粒子种群中的优质粒子,提高了算法整体适应度值,避免求解陷入局部最优解.仿真实验结果表明,基于精英策略多目标粒子群算法(ETMOPSO)提高了机械臂空间姿态的求解精度,为多关节机械臂空间姿态控制提供了一种研究思路.     

机械臂的位姿分离求逆和改进RRT-connect算法研究北大核心CSCD

针对机械臂逆解求取过程中存在大量矩阵变换、计算成本高的问题,采用位姿分离法对逆运动学求解过程进行改进,并提出基于自适应步长的RRT-connect路径规划算法。首先建立六自由度机械臂连杆坐标系模型,采用Standard Denavit-Hartenberg(D-H)方法对机械臂进行正运动学分析,得到机械臂末端执行器位姿相对于基座的齐次变换矩阵。然后引入位姿分离法改进了机械臂的逆运动学求解方法,将机械臂运动学逆解分为位置逆解和姿态逆解两部分,分别用几何法和解析法进行求解,减少了整体计算量。再者提出基于自适应步长的改进RRT-connect路径规划算法,解决了扩展速度慢的问题。最后通过仿真验证所提出方法的正确性和有效性。     

物理受限冗余机械臂逆运动学凸优化求解

针对冗余机械臂逆运动学求解结果极有可能超过机械臂物理限制的问题,提出一种基于凸优化的逆运动学求解方法使得逆解结果满足物理约束.首先分析了关节速度与力矩关系,采用机械臂动能及重复运动为优化指标,以关节速度、关节力矩为优化变量.然后将逆运动学求解问题转化为凸优化问题,进一步转化为二次规划问题,充分利用冗余特性,实现逆运动学求解时避免关节位置、关节速度、关节力矩极限.最后利用7自由度冗余机械臂KUKA LBR iiwa进行仿真,求解关节量结果符合物理极限及优化准则.结果表明本文提出的方法适用于物理受限冗余机械臂的逆运动学求解.     

基于点触任务的可变形臂逆运动学求解

为了更好地适应家庭环境下的人机交互,设计了一款可根据任务需求相应调整连杆形状的四关节可变形操作臂.与传统刚性臂相比,可变形臂具有灵巧度高、成本低和本质安全等优势,但臂形的任意性也会给操作臂运动学逆解带来额外的困难.针对可变形臂的空间点触任务,本文通过引入"点触角"这一概念放松了末端执行器的姿态约束,从而使原本欠驱动的运动学逆解问题转化为了在满足点触位置约束下,以点触角最小为优化目标的冗余臂逆解优化问题.针对该问题的求解耗时和点触精度,在旋量模型的基础上分别提出了改进的序列二次规划法(SQP)和粒子群优化(PSO)与Paden-Kahan(PSO-PK)子问题混合算法.SQP方法直接在位置层面进行非线性最优化问题求解,实验结果显示该方法在臂形由特殊变为任意时求解耗时几乎不会增加,求解效率高,对可变形臂的在线实时控制具有一定意义;同时利用分层搜索法进行初值设定可以降低该算法陷入局部极值的概率.PSO-PK算法利用Paden-Kahan子问题法的解析逆解对粒子群优化算法进行降维,实验结果表明,该算法能够在保证点触位置无误差的情况下获得稳定的最小点触角,从而使点触性能得到提高.     

机器人逆运动学差分自适应混沌粒子群求解

采用D-H法通过连杆坐标系变换矩阵建立机械臂运动控制模型,该模型呈现非常严重的非线性特性,传统方法难以求解。由于动态差分算法具有很强的全局搜索能力,而粒子群算法具有精确的局部搜索能力的特点,融合改进的动态差分算法和粒子群算法,并引入混沌映射初始种群和粒子群学习因子与惯性权重的自适应算法,提出多子群分层差分自适应混沌粒子群算法。该算法采用的多子群分层结构能提升个体共享群体信息的能力,底层利用动态差分算法进行全局搜索,顶层精英群利用改进的粒子群算法进行局部搜索。仿真试验和实际应用表明该算法在稳定性、搜索成功率以及收敛精度有显著提高,能有效解决机器人逆运动学模型的求解。     

基于BP神经网络的排爆机械臂逆运动学分析

机械臂逆运动学是已知末端执行器的位姿求解机械臂各关节变量,主要用于机械臂末端执行器的精确定位和轨迹规划,如何高效的求解机械臂运动学逆解是机械臂轨迹控制的难点;针对传统的机械臂逆运动学求解方法复杂且存在多解等问题,提出一种基于BP神经网络的机械臂逆运动学求解方法;以四自由度机械臂为研究对象,对其运动学原理进行分析,建立BP神经网络模型并对神经网络算法进行改进,最后使用MATLAB进行仿真验证;仿真结果表明:使用BP神经网络模型求解机械臂逆运动学问题设计过程简单,求解精度较高,一定程度上避免了传统方法的不足,是一种可行的机械臂逆运动学求解方法。     

空间冗余机械臂路径规划方法研究

针对空间站遥操作7DOF冗余机械臂路径规划的安全性、可靠性问题,提出了基于臂型角逆运动学的优化A*路径规划算法.本文根据臂型角参数化完善了逆运动学方法,得到了32组完备逆解集,增加了路径规划时逆解选择的灵活性;通过臂型角搜索和最小奇异值优化A*路径规划算法,提高机械臂避障、避奇异能力,机械臂操作的灵活性和路径的安全可靠性;同时根据路径优化策略,有效平滑了路径,减少了机械臂的磨损.仿真结果说明了该方法的有效性.     

基于ELM-IWO机械臂运动学逆解算法研究

为了提高机械臂控制中运动学逆解的速度和准确度,提出了一种基于入侵性杂草优化的机械臂运动学逆解方法.通过D-H法建立了工业六自由度机械臂的正向运动学模型,并利用训练速度较快的ELM(极限学习机)计算机械臂关节角度向量,即输出其逆运动学初解.利用IWO(入侵性杂草优化)算法对得到的初始逆解进行优化,取最小适应度下的杂草位置作为输出,以便得到最佳的逆运动学求解.实验结果表明,相比基于PSO-BP神经网络的求解方法,基于ELM-IWO算法的机械臂末端执行器的精度更高,实时性更好.     

基于改进天牛须算法的电力攀爬机器人运动学逆解算法

为了提高电力系统的自动化水平,减轻电力工人在检修高压输电系统时的劳动强度,同时保障电力工人人身安全,提出并设计一种可以攀爬电力铁塔的六自由度关节式机器人,针对该构型进行运动学分析和求解.为解决传统的解析法用于机械臂逆运动学求解过程中存在操作繁琐和奇异点无法逆运算等问题,提出一种基于改进天牛须算法的电力攀爬机器人运动学逆解算法.首先,对电力攀爬机器人进行DH建模,得到正运动学方程;然后,使用正运动学方程和目标位姿建立代价函数,采用改进天牛须算法对代价函数优化;最后,使用Matlab实现此算法进行仿真验证.实验结果表明,与传统的天牛须算法、改进遗传算法以及改进粒子群算法相比,所提出算法具有较好的收敛性,求解精度较高.     

Comparison of four different heuristic optimization algorithms for the inverse kinematics solution of a real 4-DOF serial robot manipulator

In this study, a 4-degree-of-freedom (DOF) serial robot manipulator was designed and developed for the pick-and-place operation of a flexible manufacturing system. The solution of the inverse kinematics equation, one of the most important parts of the control process of the manipulator, was obtained by using four different optimization algorithms: the genetic algorithm (GA), the particle swarm optimization (PSO) algorithm, the quantum particle swarm optimization (QPSO) algorithm and the gravitational search algorithm (GSA). These algorithms were tested with two different scenarios for the motion of the manipulator’s end-effector. One hundred randomly selected workspace points were defined for the first scenario, while a spline trajectory, also composed of one hundred workspace points, was used for the second. The optimization algorithms were used for solving of the inverse kinematics of the manipulator in order to successfully move the end-effector to these workspace points. The four algorithms were compared according to the execution time, the end-effector position error and the required number of generations. The results showed that the QPSO could be effectively used for the inverse kinematics solution of the developed manipulator.     

Inverse kinematics and planning/control co-design method of redundant manipulator for precision operation: Design and experiments

In this paper, we proposed two novel algorithms to improve the operating accuracy and operating efficiency of the 7-DoF redundant manipulator. Firstly, an improved adaptive particle swarm optimization (APSO) algorithm is proposed to improve the solution precision and solution speed of the inverse kinematics of the 7-DoF redundant manipulator by introducing the probability transfer mechanism and the quality evaluation criterion. Meanwhile, the velocity directional manipulability measure (VDM) is introduced as an optimization index to search for the singular-free configuration with the optimal motion performance. Then, in order to further improve the execution efficiency and stability of the 7-DoF redundant manipulator, a novel planning/control co-design (PCC) algorithm is proposed based on the Dynamic Movement Primitives (DMPs-PCC), which ensures that the motion planner and actuator of the 7-DoF redundant manipulator can work synchronously, while optimizing the velocity and acceleration profiles of each joint of the manipulator in the operating process. Finally, an experimental platform is established based on the Robot Operating System (ROS), and the effectiveness and reliability of the two novel algorithms are demonstrated by the simulations and prototype experiments.     

Cartesian path generation of robot manipulators using continuous genetic algorithms

In this paper, the authors describe a novel technique based on continuous genetic algorithms (CGAs) to solve the path generation problem for robot manipulators. We consider the following scenario: given the desired Cartesian path of the end-effector of the manipulator in a free-of-obstacles workspace, off-line smooth geometric paths in the joint space of the manipulator are obtained. The inverse kinematics problem is formulated as an optimization problem based on the concept of the minimization of the accumulative path deviation and is then solved using CGAs where smooth curves are used for representing the required geometric paths in the joint space through out the evolution process. In general, CGA uses smooth operators and avoids sharp jumps in the parameter values. This novel approach possesses several distinct advantages: first, it can be applied to any general serial manipulator with positional degrees of freedom that might not have any derived closed-form solution for its inverse kinematics. Second, to the authors’ knowledge, it is the first singularity-free path generation algorithm that can be applied at the path update rate of the manipulator. Third, extremely high accuracy can be achieved along the generated path almost similar to analytical solutions, if available. Fourth, the proposed approach can be adopted to any general serial manipulator including both nonredundant and redundant systems. Fifth, when applied on parallel computers, the real time implementation is possible due to the implicit parallel nature of genetic algorithms. The generality and efficiency of the proposed algorithm are demonstrated through simulations that include 2R and 3R planar manipulators, PUMA manipulator, and a general 6R serial manipulator.     

基于解析法和遗传算法的机械手运动学逆解

研究优化机械手轨迹规划问题,机械手运动时要具有稳定性避障性能。针对平面3自由度冗余机械手优化控制问题,建立机械手的结构模型。提出用解析法和遗传算法相结合满足具有计算量小和适应性强的特点。在给定机械手末端执行器的运动轨迹,按着机械手冗余自由度,运动轨迹上每个点对应的关节角有无穷多个解。而通过算法可以找到一组最优的关节角,可得到优化机械手运动过程中柔顺性和避障点。仿真结果表明,该算法可以快速收敛到全局最优解,可用于计算冗余机械手运动学逆解,并可实现机器人的轨迹规划和避障优化控制。     

基于自适应参数与混沌搜索的PSO算法求解柔性作业车间调度问题

针对传统粒子群优化(PSO)算法在求解柔性作业车间调度问题中的不足,提出了基于自适应参数与混沌搜索的粒子群优化算法。对粒子群算法中的惯性系数等参数采用基于迭代搜索而自适应调整的方式,使粒子在初期以较大惯性进行大范围搜索,后期逐渐减小惯性而转入精细搜索。这种方法改变了传统粒子群算法在求解过程中的盲目随机与求解精度不高的问题;同时,通过在局部搜索过程中引入混沌技术,扩大对最优解的寻找范围,以此避免算法陷入局部最优,有效提高算法的全局寻优能力。实验结果表明,基于自适应参数与混沌搜索的粒子群优化算法在求解柔性作业车间调度问题(FJSP)时能够获得更优粒子适应度平均值及更好的优化目标。所提算法对求解柔性作业车间调度问题可行,有效。     

Inverse kinematics of binary manipulators by using the continuous-variable-based optimization method

Hyper redundancy, high reliability, and high task repeatability are the main advantages of binary manipulators over conventional manipulators with continuous joints, especially when manipulators are operated under tough and complex work conditions. The precise and complex movement of a binary manipulator necessitates many modules. In this case, numerically efficient inverse kinematics algorithms for binary manipulators usually require impractically large memory size for the real-time calculation of the binary states of all joints. To overcome this limitation by developing a new inverse kinematics algorithm is the objective of this research. The key idea of the proposed method is to formulate the inverse kinematics problem of a binary manipulator as an optimization problem with real design variables, in which the real variables are forced to approach the permissible binary values corresponding to two discrete joint displacements. Using the proposed optimization method, the inverse kinematics of 3-D binary manipulators with many modules can be solved almost in real time (say, less than a second for up to 16 modules) without requiring a large memory size. Furthermore, some manipulation considerations, such as operation power minimization, can be easily incorporated into the proposed formulation. The effectiveness of the proposed method is verified through several numerical problems, including 3-D inverse kinematics problems.