一种基于场景特征的工业机器人手眼自动标定方法

A Scene Feature-based Auto Hand-Eye Calibration Method for Industrial Robot

目前,有许多方法都实现了机器人的手眼标定,但是需要借助特定的标志物,标定过程复杂,不利于工业现场的实际应用。为此,提出了一种基于场景特征的工业机器人手眼自动标定方法。首先利用光束法平差(BA)处理微运动图像序列,求解相机内参。然后利用四叉树ORB特征提取和PnP求解多视图像间的位姿变化,并使用BA优化。最后通过计算多视图像间的位姿变化和机器人的位姿变化之间的关系,获得机器人手眼位姿。实验表明,对比现有的标定算法,本文提出的算法简单快速,同时拥有较高的精度。

Nowadays,many methods have been put forward to solve the robot hand-eye calibration with the help of specific markers.The process of these calibration methods is so complex that is not suitable for practical application in industrial fields.In this paper,a scene feature-based auto hand-eye calibration for industrial robot is proposed.Firstly,the image sequence obtained by small motion is processed by bundle adjustment(BA),and the camera intrinsic parameters are solved.Then,ORB feature extraction of quadtree and PnP are used to solve pose transforms between multi-view images,followed by BA optimization.Finally,the hand-eye pose transform of the robot is obtained by calculating the relationship between the pose transforms of multi一view images and the pose transforms of the robot.Experiments show that compared with existing calibration algorithms,our algorithm is simple,fast with high accuracy.