移动机械臂运动规划研究

发布时间：2018-04-21 23:18

本文选题：移动机械臂 + 路径规划　；参考：《浙江理工大学》2017年硕士论文

【摘要】：移动机械臂是由可移动的基座和可操作的机械臂组成的复杂非线性系统。多自由度的机械臂和可移动的基座让移动机械臂兼具移动性和灵活性的双重优点,使得其在家庭服务,智能仓储和工业生产等领域中得到了广泛的应用。针对移动机械臂理论和应用的研究对于提高生活质量和促进生产力发展具有至关重要的作用,近年来已成为机器人领域的研究热点。本文针对移动机械臂运动学范畴中的若干问题进行了研究,主要内容包括:移动机械臂基座的路径规划、六自由度机械臂的逆运动学问题和机械臂避障轨迹规划的优化问题。本文的创新点主要体现在以下三个方面:(1)针对人工势场法(APF)容易产生局部极小值和路径规划效率不高的问题,提出一种基于切向量和粒子群优化的改进人工势场法(PSO-TVAPF)。首先,迭代的计算的机器人当前位置与目标之间障碍物的切向量,并通过一定的筛选策略选择最优切向量;然后,将所得切向量与传统人工势场法中的引力和斥力按照某种比例进行合成,形成机器人路径规划的驱动力,切向量的引入,对于避免局部极小值和改善路径规划质量有着显著的作用;最后,为了进一步提高算法的鲁棒性和路径规划效率,使用粒子群算法对基于切向量的人工势场法(TVAPF)进行优化。仿真实验和实物实验表明本文提出的基于切向量和粒子群优化的人工势场法能够有效的避免局部极小值和大幅度的缩短最终路径长度。(2)提出一种基于极限学习机和顺序变异的遗传算法优化的计算六自由度机械臂逆运动学解的智能算法(ELM-SGA)。算法的基本思路是先利用极限学习机求出一个精度不高的初始逆解,然后利用基于顺序变异的遗传算法优化初始逆解得到高精度解。ELM-SGA算法的提出受到基于神经网络和遗传算法求逆解的混合智能算法(Hybrid)的启发,在保证与原算法达到同等精度的情况下最大限度的提高算法的时间效率,这里的时间效率包括两个方面,神经网络的训练时间和计算逆解的时间。极限学习机随机初始化输入层权值和隐藏偏置能够最大限度提高训练速度。与传统的遗传算法的随机变异不同,本文提出一种顺序变异的方式对初步逆解进行优化,这能够有效的提高遗传算法的局部搜索能力,提高算法收敛速度。仿真实验和MT-ARM机械臂的验证也证实了本文提出的算法在保证高精度的前提下能有效提高求机械臂运动学逆解的时间效率。(3)针对机械臂的避障轨迹规划问题,提出一种改进的人工势场法,然后利用本文提出的逆解算法ELM-SGA对轨迹上的点进行求逆,进行碰撞检测,直至得到安全轨迹为止,最后利用粒子群算法以末端轨迹长度和机械臂能耗为适应度函数对轨迹进行优化。仿真实验结果表明,改进后的人工势场法能够有效缩短末端轨迹长度和降低机械臂能耗。
[Abstract]:A mobile manipulator is a complex nonlinear system composed of movable base and manipulable manipulator. The multi degree of freedom manipulator and movable base make the mobile arm with both mobility and flexibility, making it widely used in the field of home service, intelligent storage and industrial production. The research on the theory and application of the robot arm plays a vital role in improving the quality of life and promoting the development of the productive forces. In recent years, it has become a hot topic in the field of robotics. This paper has studied several problems in the kinematic category of the mobile manipulator, including the path planning of the base of the mobile manipulator, six self The inverse kinematics problem of the degree manipulator and the optimization of the trajectory planning of the manipulator obstacle avoidance are mainly embodied in the following three aspects: (1) an improved artificial potential field method based on the tangent vector and particle swarm optimization (PS) is proposed for the problem that the artificial potential method (APF) is easy to produce the local minimum and the path planning efficiency is not high. O-TVAPF). First, the iterative calculation of the current position of the robot and the tangent vector of the obstacle between the target, and select the optimal tangent vector by a certain filtering strategy. Then, the gravitational and repulsion of the tangent vector and the traditional artificial potential field method are synthesized to form the driving force of the robot path planning, and the tangent vector is formed. In the end, in order to further improve the robustness of the algorithm and the efficiency of path planning, the particle swarm optimization is used to optimize the artificial potential field (TVAPF) method based on the tangent vector. The simulation experiment and the physical experiment show that the tangent vector and particle are proposed in this paper. The artificial potential method of subgroup optimization can effectively avoid local minimum and reduce the length of the final path. (2) an intelligent algorithm (ELM-SGA) is proposed to calculate the inverse kinematics solution of the six degree of freedom manipulator based on the limit learning machine and the genetic algorithm of sequence variation. The basic idea of the algorithm is to use the limit learning machine first. An initial inverse solution with low precision is produced, and then a high precision solution.ELM-SGA algorithm is obtained by optimizing the initial inverse solution based on the genetic algorithm based on the sequence variation. It is inspired by the hybrid intelligent algorithm (Hybrid), which is based on the neural network and the genetic algorithm for the inverse solution. The time efficiency of the high algorithm, the time efficiency here includes two aspects, the training time of the neural network and the time of calculating the inverse solution. The random initialization of the input layer weight and the hidden bias of the limit learning machine can maximize the training speed. This paper proposes a sequential variation method in the random variation of the traditional genetic algorithm. The initial inverse solution is optimized, which can effectively improve the local search ability of the genetic algorithm and improve the convergence speed of the algorithm. The simulation experiment and the verification of the MT-ARM manipulator also prove that the algorithm proposed in this paper can effectively improve the time efficiency of solving the inverse kinematics of the manipulator. (3) the obstacle avoidance for the manipulator. In trajectory planning, an improved artificial potential method is proposed, and then the inverse algorithm ELM-SGA is used to reverse the points on the trajectory, and the collision detection is carried out until the security trajectory is obtained. Finally, the particle swarm algorithm is used to optimize the trajectory with the length of the end trajectory and the energy consumption of the manipulator as the fitness function. The experimental results show that the improved artificial potential field method can effectively shorten the length of the end trajectory and reduce the energy consumption of the robot arm.

【学位授予单位】：浙江理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【参考文献】