An FPGA-based Accelerator for Local Bundle Adjustment of Stereo Visual Odometry

Runzhou Guo; Man Lung Ng; Ho Tung Tang; Ray C.C. Cheung

doi:10.1109/TENCON55691.2022.9977887

An FPGA-based Accelerator for Local Bundle Adjustment of Stereo Visual Odometry

Runzhou Guo, Man Lung Ng, Ho Tung Tang, Ray C.C. Cheung

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

Abstract

This paper proposes an FPGA design and implementation of Local Bundle Adjustment (LBA) for Stereo Visual Odometry. In order to implement the LBA module on FPGA, binary Cauchy weight is designed. The LBA module is generated by Vivado HLS and implemented on the FPGA (XCZU3EG). Compared with the performance running on the FPGA and that on the Intel and ARM Processor, the LBA accelerator has achieved the highest speed which could speed up the process by 2.5 times than that on the Intel CPU, and 7.5 times than that on the ARM CPU. Based on improved LBA method on FPGA, the localization deviation is constrained within 10mm.

Original language	English
Title of host publication	2022 IEEE TENCON - Proceedings of 2022 IEEE Region 10 International Conference cum IEEE Hong Kong 50th Anniversary Celebration
Subtitle of host publication	“Tech-Biz Intelligence”
Publisher	IEEE
ISBN (Electronic)	978-1-6654-5095-9
DOIs	https://doi.org/10.1109/TENCON55691.2022.9977887
Publication status	Published - 2022
Event	2022 IEEE Region 10 International Conference, TENCON 2022 - Virtual, Online, Hong Kong, China Duration: 1 Nov 2022 → 4 Nov 2022 https://www.tencon2022.org/

Publication series

Name	IEEE Region 10 Annual International Conference, Proceedings/TENCON
Volume	2022-November
ISSN (Print)	2159-3442
ISSN (Electronic)	2159-3450

Conference

Conference	2022 IEEE Region 10 International Conference, TENCON 2022
Place	Hong Kong, China
City	Virtual, Online
Period	1/11/22 → 4/11/22
Internet address	https://www.tencon2022.org/

Research Keywords

FPGA
High Level Synthesis
Local Bundle Adjustment
Visual Odometry

Access Output

10.1109/TENCON55691.2022.9977887

Other Access Options

Check CityUHK Library

Permanent Link

Right click to copy link

Cite this

Guo, R., Ng, M. L., Tang, H. T., & Cheung, R. C. C. (2022). An FPGA-based Accelerator for Local Bundle Adjustment of Stereo Visual Odometry. In 2022 IEEE TENCON - Proceedings of 2022 IEEE Region 10 International Conference cum IEEE Hong Kong 50th Anniversary Celebration: “Tech-Biz Intelligence” (IEEE Region 10 Annual International Conference, Proceedings/TENCON; Vol. 2022-November). IEEE. https://doi.org/10.1109/TENCON55691.2022.9977887

Guo, Runzhou ; Ng, Man Lung ; Tang, Ho Tung et al. / An FPGA-based Accelerator for Local Bundle Adjustment of Stereo Visual Odometry. 2022 IEEE TENCON - Proceedings of 2022 IEEE Region 10 International Conference cum IEEE Hong Kong 50th Anniversary Celebration: “Tech-Biz Intelligence”. IEEE, 2022. (IEEE Region 10 Annual International Conference, Proceedings/TENCON).

@inproceedings{60e259821f9841ce8ba169e182b0628b,

title = "An FPGA-based Accelerator for Local Bundle Adjustment of Stereo Visual Odometry",

abstract = "This paper proposes an FPGA design and implementation of Local Bundle Adjustment (LBA) for Stereo Visual Odometry. In order to implement the LBA module on FPGA, binary Cauchy weight is designed. The LBA module is generated by Vivado HLS and implemented on the FPGA (XCZU3EG). Compared with the performance running on the FPGA and that on the Intel and ARM Processor, the LBA accelerator has achieved the highest speed which could speed up the process by 2.5 times than that on the Intel CPU, and 7.5 times than that on the ARM CPU. Based on improved LBA method on FPGA, the localization deviation is constrained within 10mm.",

keywords = "FPGA, High Level Synthesis, Local Bundle Adjustment, Visual Odometry",

author = "Runzhou Guo and Ng, {Man Lung} and Tang, {Ho Tung} and Cheung, {Ray C.C.}",

year = "2022",

doi = "10.1109/TENCON55691.2022.9977887",

language = "English",

series = "IEEE Region 10 Annual International Conference, Proceedings/TENCON",

publisher = "IEEE",

booktitle = "2022 IEEE TENCON - Proceedings of 2022 IEEE Region 10 International Conference cum IEEE Hong Kong 50th Anniversary Celebration",

address = "United States",

note = "2022 IEEE Region 10 International Conference, TENCON 2022 ; Conference date: 01-11-2022 Through 04-11-2022",

url = "https://www.tencon2022.org/",

}

Guo, R, Ng, ML, Tang, HT & Cheung, RCC 2022, An FPGA-based Accelerator for Local Bundle Adjustment of Stereo Visual Odometry. in 2022 IEEE TENCON - Proceedings of 2022 IEEE Region 10 International Conference cum IEEE Hong Kong 50th Anniversary Celebration: “Tech-Biz Intelligence”. IEEE Region 10 Annual International Conference, Proceedings/TENCON, vol. 2022-November, IEEE, 2022 IEEE Region 10 International Conference, TENCON 2022, Virtual, Online, Hong Kong, China, 1/11/22. https://doi.org/10.1109/TENCON55691.2022.9977887

An FPGA-based Accelerator for Local Bundle Adjustment of Stereo Visual Odometry. / Guo, Runzhou; Ng, Man Lung; Tang, Ho Tung et al.
2022 IEEE TENCON - Proceedings of 2022 IEEE Region 10 International Conference cum IEEE Hong Kong 50th Anniversary Celebration: “Tech-Biz Intelligence”. IEEE, 2022. (IEEE Region 10 Annual International Conference, Proceedings/TENCON; Vol. 2022-November).

Research output: Chapters, Conference Papers, Creative and Literary Works › RGC 32 - Refereed conference paper (with host publication) › peer-review

TY - GEN

T1 - An FPGA-based Accelerator for Local Bundle Adjustment of Stereo Visual Odometry

AU - Guo, Runzhou

AU - Ng, Man Lung

AU - Tang, Ho Tung

AU - Cheung, Ray C.C.

PY - 2022

Y1 - 2022

N2 - This paper proposes an FPGA design and implementation of Local Bundle Adjustment (LBA) for Stereo Visual Odometry. In order to implement the LBA module on FPGA, binary Cauchy weight is designed. The LBA module is generated by Vivado HLS and implemented on the FPGA (XCZU3EG). Compared with the performance running on the FPGA and that on the Intel and ARM Processor, the LBA accelerator has achieved the highest speed which could speed up the process by 2.5 times than that on the Intel CPU, and 7.5 times than that on the ARM CPU. Based on improved LBA method on FPGA, the localization deviation is constrained within 10mm.

AB - This paper proposes an FPGA design and implementation of Local Bundle Adjustment (LBA) for Stereo Visual Odometry. In order to implement the LBA module on FPGA, binary Cauchy weight is designed. The LBA module is generated by Vivado HLS and implemented on the FPGA (XCZU3EG). Compared with the performance running on the FPGA and that on the Intel and ARM Processor, the LBA accelerator has achieved the highest speed which could speed up the process by 2.5 times than that on the Intel CPU, and 7.5 times than that on the ARM CPU. Based on improved LBA method on FPGA, the localization deviation is constrained within 10mm.

KW - FPGA

KW - High Level Synthesis

KW - Local Bundle Adjustment

KW - Visual Odometry

UR - http://www.scopus.com/inward/record.url?scp=85145666637&partnerID=8YFLogxK

UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-85145666637&origin=recordpage

U2 - 10.1109/TENCON55691.2022.9977887

DO - 10.1109/TENCON55691.2022.9977887

M3 - RGC 32 - Refereed conference paper (with host publication)

T3 - IEEE Region 10 Annual International Conference, Proceedings/TENCON

BT - 2022 IEEE TENCON - Proceedings of 2022 IEEE Region 10 International Conference cum IEEE Hong Kong 50th Anniversary Celebration

PB - IEEE

T2 - 2022 IEEE Region 10 International Conference, TENCON 2022

Y2 - 1 November 2022 through 4 November 2022

ER -

Guo R, Ng ML, Tang HT, Cheung RCC. An FPGA-based Accelerator for Local Bundle Adjustment of Stereo Visual Odometry. In 2022 IEEE TENCON - Proceedings of 2022 IEEE Region 10 International Conference cum IEEE Hong Kong 50th Anniversary Celebration: “Tech-Biz Intelligence”. IEEE. 2022. (IEEE Region 10 Annual International Conference, Proceedings/TENCON). Epub 2022 Dec 20. doi: 10.1109/TENCON55691.2022.9977887

An FPGA-based Accelerator for Local Bundle Adjustment of Stereo Visual Odometry

Abstract

Publication series

Conference

Research Keywords

Access Output

Other Access Options

Permanent Link

Other Files and Links

Fingerprint

Cite this