基于双目视觉的机械臂位姿估计及目标自动定位系统  

Arm pose estimation and target automatic positioning system based on binocular vision

作　　者：曹阳 丁一岷 CAO Yang;DING Yimin(State Grid Jiaxing Power Supply Company,Jiaxing 314001,China)

机构地区：[1]国网嘉兴供电公司,浙江嘉兴314001

出　　处：《电子设计工程》2021年第24期107-110,115,共5页Electronic Design Engineering

摘　　要：针对以往使用的激光测距和单目摄像机估计技术、单眼照相机位姿估计法受到噪点影响而导致估计精准度低的问题,提出了基于双目视觉的机械臂位姿估计及目标自动定位系统设计。使用高压柜按钮,控制机械臂启停。使用FPGA通信协议芯片,节省信号处理时间。构建双目视觉模型,将操作目标的空间位置信息传递给机械臂,从图像中提取特征点,并求解单应性矩阵,将其转化为同一坐标系。采用基于模板匹配的粗定位目标大致位置,根据目标模板图像,识别物体位姿。利用3×3邻域高斯模板,处理图像数据,剔除噪点区域,由此得到精准位姿估计和定位结果。由实验结果可知,该系统运动坐标估计结果与预期轨迹结果基本一致,具有精准位姿估计及定位效果。In order to solve the problem of low accuracy caused by noise in laser ranging and monocular camera estimation technology and monocular camera pose estimation method used in the past,a robot arm pose estimation and an automatic target positioning system based on binocular vision were proposed.Use the high pressure cabinet knob to control the start and stop of the mechanical arm.FPGA communication protocol chip is used to save signal processing time.The binocular vision model is constructed,the spatial position information of the operating target is transmitted to the manipulator,the feature points are extracted from the image,and the homography matrix is solved,which is converted into the same coordinate system.Based on the template matching,the rough location of the target is adopted to identify the pose of the object according to the image of the target template.Gaussian template of 3×3 neighborhood was used to process image data and eliminate the area of noise points,thus obtaining accurate pose estimation and positioning results.The experimental results show that the motion coordinate estimation results of the system are basically consistent with the expected trajectory results,and have accurate pose estimation and positioning effects.

关 键 词：双目视觉 机械臂 位姿估计 目标自动定位 

分 类 号：TP29[自动化与计算机技术—检测技术与自动化装置]