Implementation of fast SLAM on a UAV

• Main Author: Muhammad Lutfan Mikail Yang Razali
• Other Authors: Xie Lihua
• Format: Final Year Project
• Language: English
• Published: 2018
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Control and instrumentation::Robotics
• Online Access: http://hdl.handle.net/10356/75276
• Institution: Nanyang Technological University

Previous Simultaneous Localization and Mapping (SLAM) methods are time-consuming iterative algorithms. A PhD student whom I am working with for this project has developed a non-iterative algorithm that produces a closed-form solution to this SLAM problem. This algorithm works with a O(n lg n) time complexity and is faster than traditional SLAM algorithms. Despite that, the non-iterative SLAM algorithm developed by the PhD student is still not efficient enough to be used optimally on Unmanned Aerial Vehicles (UAVs). This is due to the fact that UAVs require a light body for stable flight. The light weight can only be achieved by using lighter hardware. However, lighter hardware often means that they are less computationally powerful. As a result, the aim of this study was to optimize the existing non-iterative SLAM algorithm via parallel programming such that it can run optimally on incredibly low-power CPUs. This study has managed to design and restructure a non-iterative SLAM algorithm which is 34% faster than the original --- allowing the algorithm to work as efficiently on low-power CPUs as it does on computationally powerful CPUs. This allows UAVs to run the SLAM algorithm on lightweight, low-power CPUs and hence, achieve more stable flight due to the lack of heavy hardware.