仿人体3D运动的机械手臂运动轨迹自动生成的研究

发布时间：2018-09-16 20:32

【摘要】：智能化是人工智能时代的机器人技术的发展目标之一,机器人技术研究的高级目标是使得机器人可以像人一样思考、做事。如果要让机器人能够做出与人相似的智能行为,这需要机器人有很强的判断能力,以及在变化的环境中学习以获得经验知识的能力。目前的技术还不能使机器人达到这方面的要求,所以针对特定的任务,使用者需要对机器人进行编程。机器人的编程工作不仅需要专业的技术知识,而且过程复杂繁琐,费时费力,这无法达到人类对机器人给予的期望。为提高完成任务的能力,让机器人在与人的交互过程中习得新技能,本课题将研究一种使机器人通过对人体手臂运动的模仿,自动生成可执行运动轨迹的方法。与传统的通过手工编程为机器人规划轨迹的方法相比,本课题的研究将大大降低编程的难度和周期,提高机器人编程的自动化水平,增强人与机器人交互方式的智能化以及机器人的易用性,为机器人在更广阔领域的应用奠定基础。首先,本文针对如何获取人体手臂运动数据的问题,对多相机测量三维空间点坐标的原理作了介绍。通过对光学运动捕捉系统的学习和使用,实现了对运动数据的获取。其次,针对如何对机器人进行运动学建模的问题,以一种六自由度的机械臂为对象,采用D-H法建立其运动学模型,确定了机器人的D-H参数及连杆间的位姿变换矩阵,求解了正、逆运动学方程,讨论了逆解多解的取值问题,并对所求得的解进行了仿真验证。然后,针对如何描述人体手臂运动的问题,对手臂的运动进行了分析,并做相应的简化,使用少量的关节转动角度值来描述手臂的运动。在模糊逻辑理论的基础上引入了自适应神经模糊推理系统,利用神经网络的学习能力进行复杂的模糊推理,以建立复杂非线性系统的模型。最后,针对如何生成机器人可执行轨迹的问题,将获得的手臂运动数据转换为相应的机器人关节路径点,并在关节空间中,使用过关节路径点并带抛物线拟合的线性插值方法来生成机器人的运动轨迹。在相关理论算法的基础之上,编写了机器人轨迹自动生成与仿真平台,实现了对示教者运动数据的实时采集、对机器人的正确控制以及对生成轨迹的三维仿真。
[Abstract]:Intelligence is one of the development goals of robot technology in the era of artificial intelligence. The advanced goal of robot technology research is to make robots think and do things like human beings. If robots are to be able to perform intelligent behaviors similar to human beings, they need to have a strong judgment ability and the ability to learn to gain experiential knowledge in a changing environment. The current technology can not meet the requirements of the robot, so for specific tasks, the user needs to program the robot. The programming of robots requires not only professional technical knowledge, but also complicated, time-consuming and laborious processes, which cannot meet the expectations of human beings for robots. In order to improve the ability of accomplishing tasks and to acquire new skills in the process of interaction with human beings, this paper will study a method that can automatically generate executable trajectory by imitating the movement of human arm. Compared with the traditional manual programming method for robot trajectory planning, the research of this topic will greatly reduce the difficulty and cycle of programming, and improve the automation level of robot programming. The intelligentization of the interaction between human and robot and the ease of use of the robot lay the foundation for the application of robot in a wider field. Firstly, this paper introduces the principle of measuring the coordinate of three dimensional point by multi-camera aiming at how to obtain the motion data of human arm. Through the study and use of optical motion capture system, the motion data acquisition is realized. Secondly, aiming at the problem of how to model the kinematics of the robot, taking a six-degree-of-freedom manipulator as an object, the kinematics model is established by using D-H method, the D-H parameters and the position and attitude transformation matrix between the connecting rod are determined, and the positive solution is obtained. In the inverse kinematics equation, the value problem of multiple solutions of inverse solutions is discussed, and the obtained solutions are verified by simulation. Then, aiming at the problem of how to describe the movement of human arm, the motion of arm is analyzed and simplified, and the motion of arm is described by a small amount of angle of joint rotation. On the basis of fuzzy logic theory, an adaptive neural fuzzy inference system is introduced. The complex fuzzy reasoning is carried out by using the learning ability of neural network, and the model of complex nonlinear system is established. Finally, aiming at the problem of how to generate the executable trajectory of the robot, the obtained arm motion data is converted into the corresponding robot joint path point, and in the joint space, A linear interpolation method with parabola fitting is used to generate the trajectory of the robot. Based on the related theories and algorithms, a robot trajectory automatic generation and simulation platform is developed, which realizes the real-time acquisition of the motion data of the teacher, the correct control of the robot and the 3D simulation of the generated trajectory.
【学位授予单位】：华侨大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【参考文献】