Detection and handling of moving objects

Abstract

This project is to study and research on how to use robots for new applications in a catalog sale business. The catalog sale company is often faced with the problem of inefficiency in handling packages. The process of handling packages can be described as follows: first, the package is placed onto a sorting slide by a transport mechanism; at the end of the slide, the packages are then carried manually into a box on a conveyor belt. Robots can significantly reduce the cycle time by picking the packages while they are gliding. In order to do this, the robot has to know the position and orientation of the package, and the gripping region. A tracking camera that is able to track corners of the package by comparing them with the taught corner images (template matching), can determine the position and orientation of the package geometrically. The gripping region of the robot is determined by estimating the package position and orientation using a Kalman filter. As the package is approaching a gripping region, the robot will move to the package based on the information provided by the tracking camera and the Kalman filter. Using a robot not only reduces the time for the picking operation but also avoids collisions between the gripper and the package.