raullazaro/system-to-detect-and-localise-garbage-on-the-floor

Abstract

ACRO is a research group that is part of the Department of Mechanical Engineering of KU Leuven and is located in the Technology Center on the Diepenbeek campus. ACRO mainly focuses on the fields of vision and robotics and offers support to companies to integrate these technologies into their applications. One of these companies, Colruyt Group, is the largest supermarket chain in Belgium, desires to automate the cleaning processes of its stores. Nowadays, the cleaning of the floors of supermarkets is a tedious process and is currently done by human labour. First, employees have to clean the floor and remove dust and objects. Then, a scrubbing machines is employed to scrub the floor. Colruyt Group is already looking for a way to automate the scrubbing machines, but the floor is required to be free of any objects before scrubbing. The purpose of this work is to build a proof of concept of a system that detects and locates garbage on the floor of the supermarket to pick it up in a later stadium. The system should be mountable on any autonomous mobile platform that uses ROS. In this work, the TurtleBot3 has been used as a mobile platform on which a RGBD camera has been installed. For the detection of garbage, several vision methods have been integrated and compared, such as morphological operations, segmentation based on different color spaces, comparison of images and depth camera information. This project allows to demonstrate that is possible the automation of this process as well as serving as a basis for future works.