Automatic Robot-Based Unloading of Goods Out of Dynamic AGVs Within Logistic Environments

Abstract

A common and well-established technique for the transportation of goods in facility logistics is the use of Automated Guided Vehicles (AGVs). While the automated (static) loading of such transport systems is state-of-the-art, a fully automated solution for dynamically placing or unloading goods while the AGVs are moving and passing a gripper-equipped robot is still a challenge. This paper therefore proposes a new versatile and dynamic bin-picking approach for box-shaped goods using various computer vision technologies. A new method for real-time motion synchronization between a robot and an AGV will be presented, based on (1) coarse-grained AGV detection in 2D including a prediction of future positions and (2) fine-grained load detection in 3D.