基于视觉引导的机器人抓取系统手眼标定方法

Research on hand-eye calibration method of robot grasping system based on visual guidance

[目的]为降低食品自动化生产线机器人定位误差和标定物识别误差,设计一种基于空间网格的手眼标定方法。[方法]基于姿态矩阵的组成,通过平移和旋转机器人末端获取末端相对于相机的姿态与位置,提出使用已知对应点对配准的方法降低平移法中机器人定位误差对姿态求解的影响;在空间圆拟合法的基础上提出两点向量法求解末端位置以减少数据采集量,并提出拟合球心的方式补偿手眼标定结果。[结果]标定后实测平均误差为0.8 mm。[结论]基于Eye-In-Hand型手眼系统可实现较高精度的物体定位。

[Objective]Designing a apace grid-based hand-eye calibration method to reduce robot positioning errors and calibration object Recognition errors in automated food production lines.[Methods]Based on the composition of the attitude matrix,the attitude and position of the end of the robot relative to the camera were obtained by translating and rotating the end of the robot,and a registration method using known corresponding point pairs was proposed to reduce the influence of the robot positioning error on the attitude solution in the translation method.Based on the space circle fitting method,a two-point vector method was proposed to solve the end position,which reduced the amount of data collection.A method of fitting the center of the sphere was proposed to compensate for the result of hand-eye calibration.[Results]The average measured error after calibration is 0.8 mm.[Conclusion]The hand-eye system based on the Eye-In-Hand type,can realize high-precision object positioning.