Autonomous Object Pick-and-Sort Procedure for Industrial Robotics Application

Abstract

This paper describes an industrial robotics application, named Gilbreth, for autonomously picking up objects of different types from a moving conveyor belt and sorting the objects into bins according to their type. The environment, which consists of a moving conveyor belt, a break beam sensor, a 3D camera Kinect sensor, a UR10 industrial robot arm with a vacuum gripper, and different object types such as pulleys, disks, gears, and piston rods, is inspired by the NIST ARIAC competition. A first version of the Gilbreth application is implemented leveraging a number of Robot Operating System (ROS) and ROS-Industrial (ROS-I) packages. The Gazebo package is used to simulate the environment, and six external ROS nodes have been implemented to execute the required functions. Experimental measurements of CPU usage and processing times of the ROS nodes are discussed. In particular, the object recognition ROS package requires the highest processing times and offers an opportunity for designing an iterative method with the aim to fasten completion time. Its processing time is found to be on par with the time required by the robot arm to execute its movement between four poses: pick approach, pick, pick retreat and place.