双目视觉引导机器人码垛定位技术的研究

Research on Binary Vision Guided Robot Pallet Locating Technology

摘要:

针对机器人采用示教方式码垛时所遇到的误抓取等问题，提出了一种基于双目视觉引导机器人码垛的定位方法。完成了双目标定、手眼标定以及极线校正工作，利用灰度变换与图像滤波相融合的算法对图像进行预处理，提高图像的质量。将图像的灰度级分布与立体匹配算法相结合，提高立体匹配的精度，获得更好的视差图像，对视差图像进行处理得到了码垛产品区域的中心点，然后结合平行双目系统和手眼标定的结果引导机器人对产品进行定位、码垛。实验表明：该方法可实现对产品区域中心点的精确定位，获得其坐标值，引导机器人码垛。

Abstract:

Aiming at the problems of the robot's mis-grabbing when using the teaching method, the method of positioning the robot based on binocular vision is proposed. The dual target setting, hand eye calibration and polar line correction work are completed, and the image is preprocessed by the algorithm of gradation transformation and image filtering to improve the image quality. The gray level distribution of the image is combined with the stereo matching algorithm to improve the accuracy of stereo matching, obtain a better parallax image, process the parallax image to obtain the center point of the palletized product area, and then combine the parallel binocular system and the hand eye. The results of the calibration guide the robot to locate and palletize the product. Experiments show that the method can accurately locate the center point of the product area, obtain its coordinate value, and guide the robot to palletize.