基于TOF相机的四足机器人障碍物检测

发布时间：2018-04-23 22:27

本文选题：四足机器人 + TOF相机　；参考：《山东大学》2017年硕士论文

【摘要】：四足机器人与利用履带或者轮子的机器人相比不必需要连续的路径并且能够通过更为复杂的地形,并且它的运动更为灵活多变。四足机器人从机身结构来说,比腿足种类的机器人要简单的多,而且虽然结构简单,但是它们行动起来却非常的灵活稳定,在行动时四足机器人的作用机理与人类相仿,可以随机的变换自己的步伐状态来适应各种各样的地形,并且自如地前进。结合以上表述,在机器人的研究种类中,对于四足机器人的探讨研究是一个非常好的课题。随着时代的发展进步,对于机器人的各种研讨也在时刻加深,因此关于四足机器人对于环境的感知研究非常火爆。机器人的环境感知决定了它是否可以进行自主地导航,所以对于障碍物方面的检测就成为了研究的重中之重。不同于大多的其他类型的机器人,四足机器人通常活动在室外环境中,在非结构化的环境下进行一些劳作。因此本文以TOF相机为例,研究在非结构环境中,它的障碍物检测情况,下面为本文研究重点内容:(1)在绪论部分,提出了本文研究的意义与背景;其次重点介绍目前国内外对于四足机器人的学术探究情况;然后又介绍了机器人环境感知的研究现状,包括了对于环境感知的分析和其发展趋势,从而给我们构建机器人感知系统构建提供灵感;同时,我们介绍了移动机器人障碍物检测的方法;最后,给出了本文主要的研究内容等。(2)构建TOF相机、激光扫描仪等的障碍物检测系统。该系统之中,用TOF相机检测距离较近的障碍物,激光扫描仪检测距离较远的障碍物。(3)进行相机的标定。要想知道一个物体的具体三维数据,就必须对相机采取标定。这一部分主要介绍了相机的成像模型、三个坐标系的关系和运用的标定方法等,通过相机标定的实验我们求出了它的内参数矩阵。由于相机有测量范围,所以我们要对于超过相机范围的部分数据进行处理,从而减少测量时的误差。(4)在本文中,我主要向大家介绍三个类型的新型处理图像的方式。一是图像的预处理方法;二是腐蚀以及膨胀方法;三是边缘检测方法。传统图像分割方法很难用来分割非结构化环境图像,因此我们利用TOF相机能够生成灰度图像以及图像像素点三维信息的特点,通过分割图像的方式,初步地将障碍物从图像中抽离出来。最终通过边缘检测的方式检测障碍物。我们同时根据四足机器人运动特点,对于TOF相机进行了防抖动的处理,我们注意到防抖的方式有光学防抖和电子防抖,由于硬件的条件限制,TOF相机本身并不具有光学防抖,因此我们只能选择电子防抖的方法,并且对于防抖进行了实验的验证,然后结合以上的研究,得出了四足机器人可以使用该系统的可靠性结论。我们在研究的最后,关于本文的重要研究内容进行了总括并提出了未来对该项目的研究期望。
[Abstract]:Four legged robots do not need a continuous path and can pass a more complex terrain than a robot using a crawler or wheel, and its motion is more flexible. A quadruped robot is much simpler than a legged robot from the fuselage structure, and although the structure is simple, they do not act. It is often flexible and stable that the mechanism of the four legged robot is similar to the human being in action. It can adapt to all kinds of terrain and move forward at random. In the research category of the robot, the research on the quadruped robot is a very good subject. The development and progress of the robot has also been deepened at all times, so the research on the environmental perception of the quadruped robot is very hot. The environment perception of the robot determines whether it can navigate autonomously, so the detection of obstacles is the most important of the research. Type of robot, quadruped robot usually activities in the outdoor environment and do some work under unstructured environment. Therefore, this paper takes the TOF camera as an example to study the detection of its obstacles in the unstructured environment. The following is the main content of this paper: (1) the significance and background of this study are put forward in the thread theory; secondly, the importance and background of this study are put forward. This paper introduces the academic research on quadruped robots at home and abroad, and then introduces the research status of robot environmental perception, including the analysis of environmental perception and its development trend, which gives us inspiration for the construction of robot perception system. At the same time, we introduce the method of detecting obstacle in mobile robots. Finally, the main research contents of this paper are given. (2) building an obstacle detection system of TOF camera and laser scanner. In this system, a TOF camera is used to detect obstacles near the distance, and the laser scanner detects obstacles that are far away. (3) the calibration of the camera. To know the specific three-dimensional data of an object, it must be correct The camera is calibrated. This part mainly introduces the imaging model of the camera, the relation of the three coordinate systems and the calibration method of the application. By the experiment of camera calibration, we find out its internal parameter matrix. Because the camera has a measuring range, we have to deal with some of the data over the camera model, thus reducing the measurement. 4. (4) in this article, I mainly introduce three types of new image processing methods. One is the image preprocessing method; two is the corrosion and expansion method; three is the edge detection method. The traditional image segmentation method is difficult to segment the unstructured environment image, so we can use the TOF camera to generate gray scale map. As well as the features of the three-dimensional information of the pixel points of the image, the obstacles are removed from the image by dividing the images. Finally, the obstacles are detected by the edge detection. At the same time, according to the characteristics of the quadruped robot, we deal with the anti jitter of the TOF camera. We notice that the mode of anti shake is optical. Because of the restriction of hardware, the TOF camera itself does not have optical anti shake. Therefore, we can only choose the method of electronic anti shake and verify the anti shake experiment. Then, combining the above research, we get the reliability conclusion of the four legged robot which can use the system. The important research contents of this article are summarized and the future research expectation for this project is presented.

【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;TP242

【相似文献】