Seamless indoor outdoor navigation

• Main Author: Liew, Wan Seng
• Other Authors: Ang Wei Tech
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Mechanical engineering
• Online Access: https://hdl.handle.net/10356/154671
• Institution: Nanyang Technological University

Nowadays, autonomous robotic applications that operated indoor and outdoor are gaining popularity in the delivery, cleaning, and disinfection field due to the Coronavirus pandemic. Ideally, whenever there is a mobile robot applied in the real world, the robot should be capable to perform seamless indoor-outdoor navigation so that their functionalities could be more diversified. Compared to indoor navigation, which is a mostly solved topic and widely used nowadays, there is still no definite way in performing outdoor navigation, let alone travel from indoor to outdoor and vice versa. This is because the cost of applying the state-of-the-art Simultaneous Localization and Mapping (SLAM) on outdoor navigation is high and not realistic. Hence, there is a need to develop an indoor-outdoor navigation system for mobile robots to travel indoor and outdoor reliably at a reasonable cost. In this report, the navigation framework developed aims to integrate the map data extracted from the Google Maps Platform and a civil-standard GPS sensor with the conventional navigation stack of the mobile robot without utilising any SLAM element. The benefits of this framework include it eliminates the need of doing the tedious pre-mapping process and does not require any LIDAR sensor, which is commonly used in SLAM. The mobile robot platform used in this report is a wheelchair that aims to facilitate the movement of a patient with acute mobility impairment. Apart from that, this report will explore the other key components that are necessary for indoor-outdoor navigation such as the computer vision in identifying travelable outdoor paths and indoor localization sensor, Ultra-Wideband (UWB). The computer vision model utilizes the techniques of OpenCV and the performance is compared with the Nvidia Free Space Segmentation. The localization of the mobile robot is dependent on the GPS sensor outdoor while UWB at indoor, this report will also evaluate the effectiveness of both localization sensors in the navigation system.