【摘要】 介绍一种具有SLAM建图以及视觉识别功能的智能车。传统的AGV智能车通过有预设参照介质进行固定路径的运动，无法自动避障及重新规划路径，自动化程度较低，柔性较差，且存在安全隐患。为提升传统AGV智能车的自动化程度，去除了传统AGV车的巡线系统，增设了激光雷达与深度相机，通过激光雷达与深度相机的协作实现SLAM建图、路径自动规划与物体识别等功能。通过Ubuntu系统，调用功能包，向主控下达指令，Jetson Nano主控在接收到指令后，将指令下达至各电路模块，开启激光雷达和深度相机驱动。激光雷达将周围环境信息反馈到主控，由ROS系统生成实时二维地图，可自动规划路径到达指定位置。深度相机辅助定位物体，通过Open CV平台确定物体摆放姿态，并将信息传入主控系统。主控通过电机驱动来控制机械臂完成对物体的抓取工作。整个过程可在ROS机器人系统内置的RVIZ三维可视化平台中完成并监控。通过进行实地测试，智能车在完成SLAM建图，自动路径规划及识别物体并抓取这一流程上的成功率接近80%，初步掌握了该智能车正常工作的参数。更多还原

【Abstract】 A smart car with SLAM mapping and visual recognition functions is introduced. The traditional AGV intelligent vehicle moves through a fixed path with a preset reference medium,which cannot automatically avoid obstacles and re-plan the path. The degree of automation is low,the flexibility is low,and there are safety hazards. In order to improve the automation degree of traditional AGV intelligent vehicle,the patrol system of traditional AGV vehicle is removed,and the laser radar and depth camera are added. Through the cooperation of laser radar and depth camera,SLAM mapping,path automatic planning and object recognition are realized. Through the Ubuntu system,the function package is called and the instructions are issued to the main control. After receiving the instructions,the Jetson Nano main control sends the instructions to each circuit module to turn on the laser radar and depth camera drive. The lidar feeds back the surrounding environment information to the main control,and the ROS system generates a real-time two-dimensional map,which can automatically plan the path to the specified location. The depth camera assists in positioning the object,determines the pose of the object through the Open CV platform,and transmits the information to the main control system. The main controller controls the manipulator to complete the grasping work of the object through the motor drive. The whole process can be completed and monitored in the RVIZ 3D visualization platform built in the ROS robot system. Through field testing,the success rate of the smart car in completing SLAM mapping,automatic path planning and identifying objects and grasping this process is close to80%,and many parameters of the smart car’s normal operation are preliminarily mastered.更多还原