Picking and Recognizing System in Cluttered Environment

Warehouse automation is a popular research field and it has attracted significant interests of many researchers in recent years. Amazon also has intense interest in it and has held picking challenge in last three years. The achievements made during completing the challenge mission has given great insight to the research of this field. In this paper, we present a system for warehouse picking automation. To grasp object in a cluttered environment, we propose a practical method to get a reliable grasp point in a segmented object surface. And we adopt moveIt! to complete motion planning task. In recognizing process, we make a convolutional neural network to recognize grasped objects and the network is trained by our own data set.