Indoor-Positioning For Warehouse Mobile Robots Using Computer Vision

Ng, Zhun Yee (2021) Indoor-Positioning For Warehouse Mobile Robots Using Computer Vision. Final Year Project, UTAR.

Preview

PDF Download (2609Kb) | Preview

Abstract

This project is a system which is built for the warehouse. Due to the rapid growth of the e-commerce sector, the operation of the supply chain from upstream to downstream have to be more efficient and effective. In this project, the problem will be specified on warehouses/ sorting centres. The usage of automated robot machines can be easily seen in the logistics sector or supply chain processes. Every processes in a warehouse have to be effective, efficient and also in a safety condition. There is a need for the localization of the mobile robots in the indoor environment, i.e. warehouse. The indoor mobile robots are currently needed to perform various tasks. The localization feature of an indoor mobile robot is important to estimate the position and orientation of the mobile robot in the environment. Localization is required due to it is difficult for the mobile robots to perform their tasks without accurately knowing the exact position in the environment. This project can be considered as both research based and development based project. The purpose of this project is to demonstrate the usage of ceiling mounted cameras for localization of mobile robots in indoor environment. With that said, this project was developed to imitate the real world scenarios having numbers of high speed mobile robots in a huge environment. On the other hand, this project was built to determine whether the usage of overhead cameras is better than mounting the cameras on the robot itself, and to show that this project is able to use low costing equipment to perform localization of mobile robots in an indoor environment. This project proved that overhead cameras can be used to detect and track the mobile robots, which has a low costing as compared to install a camera on the robot itself. For this proposed system, a testbed based experiment will be conducted to prove that this concept works, in a much smaller scale compared to warehouse size. 2 overhead cameras with 90˚ degrees pointing down to the ground were used to capture the videos, with no overlapping of area among the cameras. To uniquely identify each of the toy car, different QR code was placed on top of the toy car for the cameras to detect. A QR code decoder was used to decode all the QR codes detected in the testbed area. All the toy cars were moved in random direction. This system can detect all the QR codes, as well as track them all along whenever detected. When the QR codes were detected, the localized position of the toy cars will be calculated and saved for later use. At the same time, the orientation of the toy car was also calculated by obtaining the coordinates of the positioning detection markers i.e. the 3 square boxes on corners of QR code. v Multiple QR codes could be present in each frame, and this proposed system could handle it. Different experiment has been conducted to get the ideal settings for the best performance of this system. This system has proven to achieve the main goal of this project, which was the localization of mobile robots in indoor environment.

Actions (login required)