ITSC 2024 Paper Abstract

Close

Paper ThAT17.11

Sun, Chaoqi (Shenzhen Institute of Advanced Technology,Chinese Academy), Xiangmin YANG, Xiangmin (Shenzhen Shenzhentong Co., Ltd), Yongfeng,Zhen, Yongfeng (Shenzhen Shenzhentong Co., Ltd), BAI, YUNHAI (Shenzhen Metro Operation Group Co.,Ltd), Peng, Lei (Shenzhen Institute of Advanced Technology,Chinese Academy of Sci)

Research on Multimodal Fusion Indoor Positioning under High-Throughput Passenger Flow : A Case Study of Metro Station

Scheduled for presentation during the Poster Session "Accurate Positioning and Localization" (ThAT17), Thursday, September 26, 2024, 10:30−12:30, Foyer

2024 IEEE 27th International Conference on Intelligent Transportation Systems (ITSC), September 24- 27, 2024, Edmonton, Canada

This information is tentative and subject to change. Compiled on October 8, 2024

Keywords Public Transportation Management, Data Mining and Data Analysis, .

Abstract

Abstract—To improve the positioning accuracy of passengers in high-throughput scenarios in the metro station, this paper proposes a multimodal fusion positioning method based on attention mechanism for UWB and vision. This method divides the metro station from the entrance to the gate into binding zones and positioning zones. Firstly a modal alignment method based on continuous trajectory matching is proposed to bind the UWB tag ID and visual data of the same passenger within the the binding zone where the passenger flow density is relatively sparse. Then, a fusion positioning model based on a hybrid attention mechanism is proposed to help estimate the accurate passenger positions within the positioning zone where the passenger flow density is higher, by fusing the matched UWB and visual positioning sequences. Experimental results show that compared with unimodal positioning methods based on UWB and vision, the positioning accuracy of this method is improved by 49.38% and 32.24%, respectively. Even compared to traditional Extended Kalman Filtering fusion positioning methods and Transformer fusion positioning models, this approach improves localization accuracy by 12.57% and 7.13%, respectively. This demonstrates the feasibility and superiority of the proposed method.

 

 

All Content © PaperCept, Inc.


This site is protected by copyright and trademark laws under US and International law.
All rights reserved. © 2002-2024 PaperCept, Inc.
Page generated 2024-10-08  13:47:43 PST  Terms of use