机器人智能抓取系统视觉模块的研究与开发

摘要

With ABB industrial robot, quick-change gripper, and binocular Camera being as hardware foun-dation, this paper develops an intelligent industrial robot grasping system based on binocular stereo vision,ai-ming at the recognition, location and automatic grasping of randomly placed objects. Based on the vision module, the principles of camera calibration, coordinate transformation, image processing and stereo matc-hing are researched, and based on C++, visual recognition and location algorithm is developed combined with OpenCV library and Triclops SDK, and the vision module can realize objects classification and pose recognition of cylinder and rectangular plate. Finally experiments are carried out to validate that the vision module can recognize cylindrical and rectangular plate objects successfully, and the positioning accuracy can meet the requirements.%以ABB机器人、快换手爪、双目相机为硬件基础,搭建了基于双目立体视觉的工业机器人智能抓取系统,旨在实现随机放置物体的识别定位和自动抓取.基于视觉模块,研究了图像处理、立体匹配和深度计算以及坐标转换,以C++为开发平台,结合OpenCV图像开发库和Triclops库,开发了视觉识别与定位算法,实现对圆柱体和长方体薄板对象的分类和位姿识别.最后,通过实验验证,表明该视觉算法有较高的识别定位精度,满足抓取系统的要求.