A 3D Vision-Based Conveyor Tracking System for Pick-and-Sort Robotic Applications

Abstract

During the process of airport baggage transfer, suitcases move continuously at high speeds and in random positions which arise considerable difficulties for automatic sorting. In order to overcome the challenges mentioned above, an automatic sorting system with a robotic arm based on 3D vision is proposed in this paper. To address shortcomings caused by the random pose of the suitcase, the rotating calipers algorithm is adopted to estimate the pose. In the dynamic picking process, a dichotomous algorithm is proposed to predict the picking position and time. Particularly in the rapid motion of the suitcases, this algorithm shows advantages of short period and higher picking accuracy. An experimental platform built at the airport to verify the algorithm, an average sorting position error of 15.17 mm and a sorting efficiency of 20 seconds per piece were achieved. Results showed that the system can accurately pick up the randomly pose suitcases which were being transferred.