CAJ | 학술논문

为增强室内移动机器人障碍物检测和道路提取能力，文中提出了一种基于深度信息的障碍检测方法。首先对深度数据进行滤波处理，填补缺失的数据；然后将深度图转换为视差图，对视差图进行水平和竖直方向投影直方图统计获得U-V视差图；由V视差图得到初步道路信息，进一步用最小二乘法拟合出完整道路平面。对U-V视差图进行两次最大类间方差法( Otsu法)分割，提取出障碍物主要信息，并根据视差关系得到障碍物在世界坐标系中的位置。实验结果表明，使用Kinect可以有效地对地面障碍物进行检测并提取出道路信息，可为室内移动机器人提供良好的导航信息。
위증강실내이동궤기인장애물검측화도로제취능력，문중제출료일충기우심도신식적장애검측방법。수선대심도수거진행려파처리，전보결실적수거；연후장심도도전환위시차도，대시차도진행수평화수직방향투영직방도통계획득U-V시차도；유V시차도득도초보도로신식，진일보용최소이승법의합출완정도로평면。대U-V시차도진행량차최대류간방차법( Otsu법)분할，제취출장애물주요신식，병근거시차관계득도장애물재세계좌표계중적위치。실험결과표명，사용Kinect가이유효지대지면장애물진행검측병제취출도로신식，가위실내이동궤기인제공량호적도항신식。
To enhance obstacle detection and road extraction ability of the indoor mobile robot,present an obstacle detection method based on depth information. Firstly,process the depth data by filtering and fill up the missing data. Then transform depth map to disparity map and calculate U-V disparity map by horizontal and vertical direction projection histogram statistics. Based on preliminary road information got by V disparity map,can fit a complete road plane by using the least square method. With twice segmentation of the U-V disparity map by Otsu’ s method,extract the main information of obstacles and obtain the location of obstacle in the world coordinate according to disparity relationship. Experiments show that Kinect can effectively improve the ability of obstacle detection and road information extrac-tion for indoor mobile robot,and provide good navigation information.