| Author: | Liu, Tianxia |
| Title: | GNSS/UWB/VIO integration for precise and seamless robotic localization |
| Advisors: | Chen, Wu (LSGI) |
| Degree: | Ph.D. |
| Year: | 2024 |
| Subject: | Robots -- Control Mobile robots Robots -- Control systems Hong Kong Polytechnic University -- Dissertations |
| Department: | Department of Land Surveying and Geo-Informatics |
| Pages: | xvii, 202 pages : color illustrations |
| Language: | English |
| Abstract: | Autonomous mobile robots are becoming increasingly prominent in modern societies, revolutionizing various industries with their wide-ranging applications. To ensure precise autonomous operations, seamless localization with centimeter to decimeter level accuracy is required. However, the complex urban environments and the equipment of low-cost devices pose significant challenges in achieving reliable and accurate robot localization. In this thesis, a multi-sensor integrated system combining low-cost GNSS, UWB, MEMS IMU, and stereo cameras is developed to enhance robot positioning in complex and seamless environments. The core idea is to first improve absolute positioning accuracy and continuity in indoor and outdoor environments, thus ensuring precise absolute references. Then, incorporate relative positioning sensors to enhance positioning in highly obscured areas. To improve UWB positioning accuracy in large-scale indoor scenarios, two innovative techniques for systematic error mitigation are proposed. Firstly, an antenna phase center offset (PCO) calibration model is established with respect to the tag-to-anchor elevation angles to reduce the PCO effects in UWB range measurements. Secondly, to mitigate the unmodeled environmental errors introduced by time latencies calibrated at a single point, a multi-point time latency determination (MTLD) method is proposed to provide precise time latencies over a large area. The results show that the established PCO calibration model can correct the PCO errors by around 70% in UWB range measurements. The accuracy of time latency determination is improved by 10% with MTLD, and further by 44% with both MTLD and PCO calibration. To improve GNSS positioning continuity in urban environments, a cycle slip estimation method with position polynomial fitting (PPF) constraint is proposed. It is dedicated to avoiding the frequent ambiguity re-fixing in RTK positioning through efficient cycle slip repair. As for applications using low-cost GNSS receivers, the conventional measurement polynomial fitting (MPF) and triple-differenced residual-based snooping (TRS) methods fail in continuous and multiple cycle slips cases, while the geometry-based (GB) method exhibits low cycle slip repair success rates. Motivated by the principle that a more accurate prior baseline contributes more efficient cycle slip estimation, the new method imposes a precisely formed positional constraint into GB model to enhance cycle slip estimation. The results show that PPF improves the success rates of cycle slip repair and ambiguity resolution by 25.7% and 62.4%, 40.1% and 46.5%, respectively, compared with TRS and MPF. To further enhance GNSS cycle slip processing in highly obscured environments, a VIO-aided cycle slip estimation method is proposed. Considering that VIO provides accurate relative pose with less drifts, the new method introduces VIO-predicted baseline constraint into the GB model to further enhance cycle slip estimation. The method is implemented in a TC algorithm of GNSS RTK and VIO integration. The results show that it can repair cycle slips after 10s∼60s GNSS data gaps, in which PPF and MPF are unavailable. Compared with the measurement-based MVIO, and the INS-aided MINS and GINS, the positioning accuracy of the proposed algorithm is improved by 26.2%, 65.9%, and 57.4%, respectively. To enhance seamless urban positioning, a tightly coupled (TC) GNSS/UWB/VIO integration system is developed. An EKF-based TC framework including the functions of system initialization, decentralized filter updating and quality control is designed. A synchronous platform and a ROS package are developed for multi-sensor data collection and processing. Real-world tests show that the developed system achieves 0.411m and 0.077m horizontal positioning in complex outdoor and indoor environments, yielding 71.2% and 18.1% improvements compared with the traditional loosely coupled schemes, which fulfills the requirements of precise and seamless robot positioning. |
| Rights: | All rights reserved |
| Access: | open access |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/13091