基于RGB-D传感器的车间三维场景建模  

Three-Dimension Modeling of Manufacturing Plant Scene Based on RGB-D Sensor

作　　者：全思博[1] 李伟光[1] 郑少华[1] 

机构地区：[1]华南理工大学机械与汽车工程学院,广东广州510640

出　　处：《华南理工大学学报（自然科学版）》2015年第6期42-47,共6页Journal of South China University of Technology(Natural Science Edition)

基　　金：广东省数控一代机械产品创新应用示范工程专项(2013B011301026);广东省教育部产学研结合项目(2012B091100311)~~

摘　　要：感知车间环境是实现自动引导小车(AGV)智能化自主运行的基础.三维场景建模比二维地图更适于表达非结构化场景信息,然而已有的激光雷达技术和视觉技术实现方法都有其不足之处.文中利用RGB-D深度传感器,结合移动机器人及开源机器人操作系统(ROS)建构了一套软硬件系统,用于制造车间环境的复杂大场景三维建模;介绍了该系统光机电硬件和ROS软件的各模块功能与集成设计,以及相应的扫描策略,并用该集成系统进行了车间局部环境自动连续扫描、数据拼接、三维建模实验.结果表明用该系统可以较好地完成三维场景建模任务.Environment perception is a prerequisite for the autonomous intelligent operation of automatic guided ve- hicles, and 3D maps are more suitable than the 2D ones for representing an unstructured scene. However, the exi- sting laser radar- and vision-based methods both have some deficiencies. In this paper, a RGB-D sensor mounted on a mobile robot with open-source ROS framework is used to build an integrated system for the three-dimension modeling of large-scale and complex scenes in manufacturing plants, and the architecture of hardware and ROS soft- ware modules of the system as well as the scanning strategy is introduced. Moreover, an experiment is carried out to investigate the automatic continuous scanning of a region in a manufacturing plant scene, the data rezistration and the three-dimension modeling of the system. The results show that the proposed system meets the requirements of three-dimension scene modeling well.

关 键 词：移动机器人 三维场景建模 ROS 车间环境 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]