基于双目视觉障碍物检测的翻车保护装置自动折叠系统
发布时间：2018-11-22 19:13
【摘要】：随着人们对坐骑式割草机车辆在大型草坪场所中对割草效率和车辆行驶安全性的要求越来越高,坐骑式割草机车辆越来越智能化,安全措施也逐步提高。其中安装在车辆上的翻车保护装置便是起到割草机车辆意外侧翻时保护驾驶员安全的作用。但是其翻车保护装置因安全设计需求结构设计偏高,当坐骑式割草机进行割草作业时,会与作业场所中的半空的树枝等静态障碍物发生碰撞干涉,影响坐骑式割草机的割草效率,甚至会因碰撞导致车辆侧翻,危机驾驶员的生命安全,也违背了设计该翻车保护装置的初衷,因此对基于双目视觉障碍物检测的翻车保护装置自动折叠系统的研究设计是一件很有实际意义的工作。本研究设计从两方面着手,一对原翻车保护装置进行硬件改进,使翻车保护装置上保护杠可被自由控制折叠。其二添加障碍物实时检测系统装置,实时检测在车辆行驶时前方出现的障碍物,并判断是否将与翻车保护装置发生碰撞并干涉车辆正常行驶。原翻车保护装置改进:原翻车保护装置的上下保护杠采用插销固定连接,无法被实时控制折叠,为此对其进行改进,上下保护杠采用减速机进行自由链接,并添加了PLC电气控制装置,可由此实时控制上保护杠进行折叠,以降低翻车保护装置的整体高度,从而可避免与半空中的障碍物发生碰撞干涉。障碍物检测系统:采用基于双目视觉的障碍物检测,其基本原理是利用障碍物在双目视觉中产生的双目视差图与障碍物三维属性之间的线性对应关系,进行障碍物的三维属性判断。这种检测方法可适用于外界复杂环境背景中无明显颜色和边界特征的静态障碍物检测,具体检测流程包括双目摄像机标定、双目图像获取、双目校正、立体匹配、获取双目视差图和利用双目视差图进行障碍物分析过程。本障碍物检测进行障碍物三个方面检测包括:(1)障碍物与翻车保护装置的深度距离信息分析,判断翻车保护装置与障碍物发生碰撞的时间性;(2)障碍物的大小粗度信息分析,判断障碍物是否具备阻碍坐骑式割草机继续车辆行驶的能力;(3)障碍物的阻挡区域信息分析,判断是否具备与翻车保护装置发生碰撞的可能性。系统依据以上三方面结果实时控制翻车保护装置进行自动折叠,以此提高翻车保护装置对驾驶员的安全保护性能。
[Abstract]:With the increasing demands on the efficiency of lawn mower and the driving safety of the vehicle in the large lawn, the riding lawn mower vehicle becomes more intelligent and the safety measures are improved step by step. The overturning protection device installed on the vehicle is used to protect the driver's safety when the mower vehicle overturns accidentally. However, due to the high structural design requirements for safety design, the overturning protection device will collide with static obstacles such as half-empty branches in the working place when the riding lawn mower is mowing. The impact on the efficiency of the riding mower will even cause vehicle rollover caused by collision, which will endanger the life safety of the driver, which also violates the original intention of designing the protection device for the vehicle. Therefore, the research and design of automatic folding system based on binocular visual obstacle detection is of great practical significance. The design of this study is based on two aspects. A pair of original overturning protection devices are improved in hardware so that the protection bars on the overturn protection device can be freely controlled and folded. The second is to add the obstacle real-time detection system device to detect the obstacle in front of the vehicle, and to judge whether the vehicle will collide with the overturning protection device and interfere with the normal running of the vehicle. The original overturning protection device is improved: the upper and lower protection bars of the original overturn protection device are fixed by a pin, which cannot be folded in real time, so it is improved, and the upper and lower protection bars are freely linked by a reducer. The PLC electric control device is added, which can fold the upper protection bar in real time, so as to reduce the overall height of the overturning protection device and avoid collision interference with obstacles in mid-air. Obstacle detection system: the basic principle of obstacle detection based on binocular vision is to use the linear correspondence between the binocular parallax generated by obstacle in binocular vision and the three-dimensional attributes of obstacle. The three-dimensional attribute of obstacles is judged. This detection method can be applied to the detection of static obstacles without obvious color and boundary characteristics in complex environment. The detection flow includes binocular camera calibration, binocular image acquisition, binocular correction, stereo matching, etc. The binocular parallax map is obtained and the obstacle analysis process is carried out by using the binocular parallax map. The obstacle detection includes the following three aspects: (1) analyzing the depth distance information between the obstacle and the overturning protection device to judge the time of collision between the overturn protection device and the obstacle; (2) analyzing the information of the size and roughness of the obstacle, judging whether the obstacle has the ability to prevent the riding lawn mower from moving on; (3) analyzing the blocking area information of obstacles and judging whether there is the possibility of collision with the overturning protection device. According to the above three results, the system can control the automatic folding of the overturning protection device in real time, so as to improve the safety protection performance of the overturning protection device to the driver.
【学位授予单位】：江苏科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TU986.32;TP391.41
[Abstract]:With the increasing demands on the efficiency of lawn mower and the driving safety of the vehicle in the large lawn, the riding lawn mower vehicle becomes more intelligent and the safety measures are improved step by step. The overturning protection device installed on the vehicle is used to protect the driver's safety when the mower vehicle overturns accidentally. However, due to the high structural design requirements for safety design, the overturning protection device will collide with static obstacles such as half-empty branches in the working place when the riding lawn mower is mowing. The impact on the efficiency of the riding mower will even cause vehicle rollover caused by collision, which will endanger the life safety of the driver, which also violates the original intention of designing the protection device for the vehicle. Therefore, the research and design of automatic folding system based on binocular visual obstacle detection is of great practical significance. The design of this study is based on two aspects. A pair of original overturning protection devices are improved in hardware so that the protection bars on the overturn protection device can be freely controlled and folded. The second is to add the obstacle real-time detection system device to detect the obstacle in front of the vehicle, and to judge whether the vehicle will collide with the overturning protection device and interfere with the normal running of the vehicle. The original overturning protection device is improved: the upper and lower protection bars of the original overturn protection device are fixed by a pin, which cannot be folded in real time, so it is improved, and the upper and lower protection bars are freely linked by a reducer. The PLC electric control device is added, which can fold the upper protection bar in real time, so as to reduce the overall height of the overturning protection device and avoid collision interference with obstacles in mid-air. Obstacle detection system: the basic principle of obstacle detection based on binocular vision is to use the linear correspondence between the binocular parallax generated by obstacle in binocular vision and the three-dimensional attributes of obstacle. The three-dimensional attribute of obstacles is judged. This detection method can be applied to the detection of static obstacles without obvious color and boundary characteristics in complex environment. The detection flow includes binocular camera calibration, binocular image acquisition, binocular correction, stereo matching, etc. The binocular parallax map is obtained and the obstacle analysis process is carried out by using the binocular parallax map. The obstacle detection includes the following three aspects: (1) analyzing the depth distance information between the obstacle and the overturning protection device to judge the time of collision between the overturn protection device and the obstacle; (2) analyzing the information of the size and roughness of the obstacle, judging whether the obstacle has the ability to prevent the riding lawn mower from moving on; (3) analyzing the blocking area information of obstacles and judging whether there is the possibility of collision with the overturning protection device. According to the above three results, the system can control the automatic folding of the overturning protection device in real time, so as to improve the safety protection performance of the overturning protection device to the driver.
【学位授予单位】：江苏科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TU986.32;TP391.41
【参考文献】
相关期刊论文 前10条
1 郭文俊;;基于三帧间差分的障碍物检测算法[J];工业控制计算机;2016年01期
2 冯亮;谢劲松;李根;霍庆立;;摄像机标定的原理与方法综述[J];机械工程师;2016年01期
3 黄应清;文军;谢志宏;;摄像机畸变的非量测校正方法研究[J];现代电子技术;2015年20期
4 杨欢;;一种基于视觉避障及导航功能的机器人设计[J];机电技术;2015年03期
5 姬长英;沈子尧;顾宝兴;田光兆;张杰;;基于点云图的农业导航中障碍物检测方法[J];农业工程学报;2015年07期
6 潘峥嵘;马莉;;基于特征信息提取的双目视觉障碍物检测研究[J];计算机测量与控制;2013年07期
7 晁衍凯;徐昱琳;周勇飞;吕晓梦;王明;;基于双目视觉的机器人目标定位与机械臂控制[J];计算机技术与发展;2013年07期
8 胡天翔;郑加强;周宏平;;基于双目视觉的树木图像测距方法[J];农业机械学报;2010年11期
9 刘振中;傅莉;;摄像机标定研究[J];沈阳航空工业学院学报;2010年01期
10 沈志熙;黄瀚敏;黄席樾;李敏;;基于视觉的室外移动机器人障碍物检测方法[J];机器人;2009年04期
相关硕士学位论文 前1条
1 张e，
本文编号：2350284
本文链接：https://www.wllwen.com/jianzhugongchenglunwen/2350284.html
