基于双卷积链的贴标机械手产品中心视觉定位方法OA

At present,the passive binocular stereo vision positioning method is mainly used for the positioning of the visual center of robotic arm products.However,due to the lack of scene coordinate matching and prediction,its visual positioning accuracy is poor.Therefore,a visual center positioning method for labeling robotic arm products based on dual convolutional chains is proposed.Firstly,the built-in and external parameters of the camera is calibrated to construct a visual positioning scene coordinate system.Then,using the dual convolutional network structure in convolutional neural networks,the visual center scene coordinates are predicted.Finally,the visual positioning of the labeling robot product center is achieved by matching pixel coordinates with image feature points.In the experiment,the positioning accuracy of the proposed method is tested in a 40 mm×40 mm×40 mm small-scale space scene and 180 mm×180 mm×180 mm.The experimental results show that when using the proposed method for visual positioning of the center of the labeling robotic arm product,the positioning variance is small and has relatively ideal positioning accuracy.The proposed method has good application prospects in the positioning field of robotic arm products,which helps to achieve industrial automation production.