基于视觉的驾驶机器人导航技术  

Navigation Technology of Driving Robot Based on Machine Vision

作　　者：李旭[1] 张为公[1] 

机构地区：[1]东南大学仪器科学与工程学院,南京210096

出　　处：《光电工程》2008年第2期56-60,共5页Opto-Electronic Engineering

基　　金：江苏省汽车工程重点实验室开放基金资助项目(QC200603);江苏省交通科学研究计划项目(06C04);苏州市科技发展计划(工业攻关)项目(SG0723);东南大学博士科研启动基金(4022001008)

摘　　要：对驾驶机器人的视觉导航技术进行了研究,提出了一种快速可靠的车辆引导线检测算法。该算法首先对感兴趣的移动窗口进行有条件的边缘检测,进而引入了梯度方向直方图的处理思想,制定了车辆引导线边缘候选点集与有效点集的筛选算法,最后,通过所提出的一种基于标量化处理Kalman滤波方法实现了引导线特征参数的快速准确提取。试验表明,本文提出的检测算法能够满足复杂路况下的导航需要,具有30ms/帧的实时处理速度。The navigation technology based on machine vision for driving robot in vehicle road tests is studied, A quick and robust image algorithm to detect vehicle inductive lines is proposed. In this algorithm, the conditional edge detection is carried out firstly within an interested mobile region. Then, a novel processing idea of gradient direction histogram is introduced. And how to select Candidate point sets and valid point sets on the inductive line edges is ascertained. Finally, the feature parameters about vehicle inductive lines are extracted quickly and accurately by the proposed Kalman filter with the scalar measurement processing. Experiments show that the proposed detection algorithm can satisfy the navigation demands of the driving robot even in the complex road environment and has the real-time processing velocity as quick as 30ms per frame.

关 键 词：驾驶机器人 机器视觉 导航 引导线检测 

分 类 号：U666.1[交通运输工程—船舶及航道工程]