Design of Industrial Robot Sorting System with Visual Guidance Based on Webots

Abstract

Aiming at the low efficiency of artificial detection in industrial production line, this paper builds a simulation platform of visual guide sorting system based on Webots. For static target work pieces, image is captured by the camera, use Hu invariant moments quick recognition, and get the help of chain code to further identify, so that the sorting task can be done by robots. For dynamic target work pieces, it takes much time to process the image. There will be a tracking lag phenomenon. Using Kalman filter to predict the position, we can track the target. Experimental results showed that, with the feature in the centroid of shape and the tracking error in two pixels or less, the system could achieve target-detecting and tracking at the 0.2Sm/s speed of conveyor, as well as have a better robustness.