基于Sylvester方程变形的荔枝采摘机器人手眼标定方法

发布时间：2018-06-29 10:01

本文选题：机器人 + 标定　；参考：《农业工程学报》2017年04期

【摘要】：针对视觉荔枝采摘机器人的Eye-in-Hand视觉与机器人关联方式的手眼标定问题,该文提出一种基于优化的求解齐次变换矩阵方程的手眼标定方法。该方法通过机器人带动其臂上的双目相机从多个位置观测标定板,使用Sylvester方程变形对手眼标定近似方程线性化,再对简单的初值进行优化计算,最终得到精确的标定结果。该方法的软件用C++/Open CV开发实现,并进行了多个试验。试验结果表明,视觉与机器人关联后,定位误差与机器人运动次数相关,当距目标1 m左右,静态时的视觉系统误差均值为0.55 mm;动态工作时,视觉关联机器人重复定位误差的均值为2.93 mm,标准差为0.45 mm,符合具有容错功能的视觉荔枝采摘机器人的实际使用需求。使用基于Sylvester方程变形的手眼标定方法标定的视觉荔枝采摘机器人,在野外环境下,总体采摘成功率达到76.5%,视觉系统成功识别、定位采摘点的情况下,采摘成功率达92.3%。
[Abstract]:Aiming at the hand-eye calibration problem of Eye-in-Hand vision and robot correlation mode of visual litchi picking robot, this paper presents a hand-eye calibration method based on optimization to solve homogeneous transformation matrix equation. This method uses the robot to drive the binocular camera on its arm to observe the calibration board from multiple positions, linearizes the approximate equation of the eye calibration using Sylvester equation, then optimizes the simple initial value, and finally obtains the accurate calibration results. The software of this method is developed with C / Open CV and many experiments are carried out. The experimental results show that the positioning error is related to the number of robot movements after the vision is associated with the robot. When the distance from the target is about 1 m, the mean value of the visual system error is 0.55 mm in static state, and the mean value is 0.55 mm when the robot is moving. The mean value of repeat location error and standard deviation of visual correlation robot are 2.93 mm and 0.45 mm respectively, which accord with the practical application requirement of visual litchi picking robot with fault-tolerant function. The visual litchi picking robot calibrated by hand eye calibration method based on Sylvester equation was used. In the field environment, the overall success rate of picking was 76.5. the successful rate of picking was 92.392.3when the vision system was successful in identifying and locating picking points.
【作者单位】：华南农业大学南方农业机械与装备关键技术重点实验室;
【基金】：国家自然科学基金资助项目(31571568) 广东省科技项目(2015A020209111) 广州市科技计划(201510010140)
【分类号】：TP391.41;TP242

【相似文献】