多约束下的机械臂运动控制算法研究

发布时间：2018-06-20 08:00

本文选题：机械臂 + 运动学控制　；参考：《浙江大学》2016年博士论文

【摘要】：随着机器人行业的快速发展,机械臂已经深入到工业制造的各个方面,并且往医疗以及家庭服务方向迈进。针对应用的不同,机械臂的结构类型多样化,关节数从两关节到七关节不等,负载以及速度运行能力都有不同的等级。因而,研究通用的运动控制算法,适用于不同种类的机械臂控制至关重要。另一方面,工业现场总线的通讯速率已经可以满足机械臂的实时运动控制要求,而且计算机运行速度也可以保证复杂算法在控制周期内完成。因而,机械臂开始从原先的离线规划转变到实时控制模式,对运动控制算法的鲁棒性提出了更高要求。此外,机器人的运行环境已不再单一,在复杂工况以及动态环境中,机械臂运动控制算法必须具备处理多任务、多约束的能力。本文以实验室自主研制的七关节乒乓球机器人为载体,通过理论研究和实验仿真验证相结合的方式,系统深入地研究了机械臂姿态运动控制、关节约束下的运动控制、多任务优先级运动控制以及多约束下的运动控制等问题,相关成果应用于工业机器人通用控制软件平台中,并且为机器人在复杂环境中的应用提供有效的理论支持和技术方案。论文的主要研究内容如下:1)针对某些机械臂应用场景中,姿态定向可以实现与定位等同的功能,本文提出了机械臂的姿态定向控制方法。通过引入四元数法,实现了姿态大范围变化的连续参数表达。根据姿态定向特性,提出了误差反馈形式,并且采用李雅普诺夫方法,证明了定向控制闭环系统的渐进稳定特性。采用姿态定向控制,相对于定位任务,机械臂多了一个冗余度,从而提升机械臂的操作度,以及关节避障的能力。2)提出了关节约束下机械臂的运动学控制算法,钳位加权最小范数算法。由于加权最小范数法的权值切换条件需要对关节速度进行过零检测,因而容易受到干扰而发生错误的权值切换判断,导致关节速度指令不连续。本文将关节钳位项引入到加权最小范数算法中,提出了钳位加权最小范数算法。该算法采用连续权值,消除了原先的权值切换问题,并且引入钳位任务,有效地防止关节运行至限位之外,促使关节远离限位。钳位加权最小范数法适用于非冗余以及冗余机械臂。在主任务与关节限位不发生冲突的情况下,该方法可以确保主任务的精度。3)任务优先级的控制算法是处理多任务的有效方法。然而,任务的有效性切换将会引起关节速度的不连续。本文提出一种新的零空间计算方式,并且给出了严格的数学证明。在此基础上,提出连续的零空间算子以及矩阵逆计算,应用于优先级任务分层控制中,实现任务的有效性切换,并且保证关节速度的连续性。4)广义加权最小范数法中的虚拟关节将一般约束问题转化为关节约束问题处理。然而,虚拟关节转换矩阵存在奇异问题,以及虚拟关节必须小于实际关节个数。另外,将多个约束转化为一个虚拟关节,且不允许多个约束同时触发,在实际中是难以满足的。本文采用新的方式构建虚拟关节,解决了上述问题,并且将虚拟关节技术应用于梯度投影法,加权最小范数法以及钳位加权最小范数法中。
[Abstract]:With the rapid development of the robotics industry, the manipulator has penetrated into all aspects of industrial manufacturing and moved into the direction of medical and family services. According to the different applications, the structure of the manipulator is diversified, the number of joints varies from two joints to seven joints, and the capacity of load and speed is different. The motion control algorithm used is very important for different kinds of manipulator control. On the other hand, the communication rate of the industrial fieldbus can satisfy the real-time motion control requirements of the manipulator, and the computer running speed can also ensure that the complex algorithm is completed within the control cycle. Therefore, the manipulator begins to go from the original off-line. In addition, the operating environment of the robot is no longer single. In complex working conditions and dynamic environment, the manipulator motion control algorithm must have the ability to deal with multi task and multi constraint in the complex working conditions and dynamic environment. The seven joint table tennis developed by the laboratory is developed in this paper. The robot is a carrier. Through the combination of theoretical research and experimental simulation verification, the system has deeply studied the control of the attitude motion of the manipulator, the motion control under the joint constraints, the multi task priority motion control and the motion control under multiple constraints. The related achievements are applied to the general control software platform of industrial robots. And it provides effective theoretical support and technical scheme for the application of robot in complex environment. The main research contents of this paper are as follows: 1) in the application scene of some manipulator, the attitude orientation can be equal to the positioning. In this paper, the attitude direction control method of the manipulator is proposed. The four element method is introduced to realize the posture. According to the attitude orientation characteristic, the error feedback form is proposed, and the asymptotic stability characteristic of the directional control closed loop system is proved by the Lyapunov method. The attitude orientation control is used to improve the manipulator's operating degree by a redundancy relative to the positioning task. And the ability of joint obstacle avoidance (.2), the kinematics control algorithm and the clamp weighted minimum norm algorithm under the joint constraint are proposed. The weight switching condition of the weighted minimum norm method requires the zero crossing detection of the joint velocity, which is easy to be judged by the right value switching of the error, resulting in the discontinuity of the joint velocity instruction. In this paper, the joint forceps are introduced into the weighted minimum norm algorithm, and the forceps weighted minimum norm algorithm is proposed. The algorithm uses continuous weights to eliminate the original weight switching problem, and introduces the clamp task to effectively prevent the joint from running to the limit, so that the joint is far away from the limit. Redundancy and redundant manipulator. In the case of no conflict between the main task and the joint limit, the method can ensure the accuracy of the main task.3) the task priority control algorithm is an effective method to deal with the multitask. However, the task's effectiveness switching will cause the discontinuity of the joint speed. A new zero space computing party is proposed in this paper. On the basis of this, a continuous zero space operator and inverse matrix calculation are proposed. In the hierarchical control of priority tasks, the effectiveness of the task is switched, and the continuity of the joint velocity is guaranteed. The virtual joint in the generalized weighted minimum norm method transforms the general constraint problem into the.4. However, the virtual joint transformation matrix has a singular problem, and the virtual joint must be less than the number of actual joints. In addition, multiple constraints are converted into a virtual joint, and many constraints are not allowed to trigger at the same time. In this paper, a new method is used to construct virtual joints and solve the above mentioned above. The virtual joint technology is applied to the gradient projection method, the weighted minimum norm method and the clamping weighted minimum norm method.
【学位授予单位】：浙江大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TP242

【参考文献】