Low–cost visual localization system on an embedded platform

• Main Author: Sua, Heng Duang
• Other Authors: Lam Siew Kei
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Computer science and engineering::Hardware::Performance and reliability
• Online Access: https://hdl.handle.net/10356/157237
• Institution: Nanyang Technological University

Simultaneous Localization and Mapping (SLAM) of an unknown environment is very crucial in navigation for autonomous robots. The role of SLAM is more critical for indoor navigation as the SLAM system cannot rely on Global Positioning System (GPS). Furthermore, an autonomous robot is typically only equipped with camera sensors and a low-cost embedded system, which poses a challenge in achieving high-speed visual SLAM. The objective of this project is to develop a preliminary Field Programmable Gate Array (FPGA) based sensing and computing stack for visual SLAM. To date, most of the existing visual SLAM algorithms have been implemented on microprocessors and GPUs. This project aims to port a widely used visual-inertial SLAM framework to the processing system (PS) of an FPGA platform and perform calibration of the Inertial Measurement Unit (IMU) and camera sensors to reduce pose estimate uncertainty. This project will lay the foundation for future research in hardware acceleration of visual SLAM algorithm on FPGA