A method, system and device for identifying a shelf body of a shelf

Abstract

The application provides a shelf body recognition method, system and device, and relates to the technical field of image processing.The recognition method comprises the following steps: the real-time visual recognition result (column number recognition result data) is dynamically fused and calibrated with SLAM navigation data, so that the problem of SLAM static map failure caused by the position change of the dense shelf in the prior art can be effectively overcome; the method does not need to rely on fixed environment structure, nor does it need to additionally arrange auxiliary marks, only the inherent column number of the shelf body is used as the visual calibration basis, a light-weight target detection model is combined with a pre-constructed mapping library, image acquisition, column number recognition, coordinate matching and deviation correction are completed in real time during the movement of the robot, the positioning accuracy and operation reliability of the robot in the dynamic environment of the dense shelf are significantly improved, meanwhile, the method has strong environmental adaptability (such as anti-light change and dust interference) and system compatibility, and the deployment and maintenance cost is reduced.

Images