Event3DGS: Event-Based 3D Gaussian Splatting for High-Speed Robot Egomotion

Tianyi Xiong, Jiayi Wu, Botao He, Cornelia Fermuller, Yiannis Aloimonos, Heng Huang, Christopher Metzler
Proceedings of The 8th Conference on Robot Learning, PMLR 270:4100-4118, 2025.

Abstract

By combining differentiable rendering with explicit point-based scene representations, 3D Gaussian Splatting (3DGS) has demonstrated breakthrough 3D reconstruction capabilities. However, to date 3DGS has had limited impact on robotics, where high-speed egomotion is pervasive: Egomotion introduces motion blur and leads to artifacts in existing frame-based 3DGS reconstruction methods. To address this challenge, we introduce Event3DGS, an event-based 3DGS framework. By exploiting the exceptional temporal resolution of event cameras, Event3GDS can reconstruct high-fidelity 3D structure and appearance under high-speed egomotion. Extensive experiments on multiple synthetic and real-world datasets demonstrate the superiority of Event3DGS compared with existing event-based dense 3D scene reconstruction frameworks; Event3DGS substantially improves reconstruction quality (+3dB) while reducing computational costs by 95%. Our framework also allows one to incorporate a few motion-blurred frame-based measurements into the reconstruction process to further improve appearance fidelity without loss of structural accuracy.

Cite this Paper

BibTeX
@InProceedings{pmlr-v270-xiong25b, title = {Event3DGS: Event-Based 3D Gaussian Splatting for High-Speed Robot Egomotion}, author = {Xiong, Tianyi and Wu, Jiayi and He, Botao and Fermuller, Cornelia and Aloimonos, Yiannis and Huang, Heng and Metzler, Christopher}, booktitle = {Proceedings of The 8th Conference on Robot Learning}, pages = {4100--4118}, year = {2025}, editor = {Agrawal, Pulkit and Kroemer, Oliver and Burgard, Wolfram}, volume = {270}, series = {Proceedings of Machine Learning Research}, month = {06--09 Nov}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v270/main/assets/xiong25b/xiong25b.pdf}, url = {https://proceedings.mlr.press/v270/xiong25b.html}, abstract = {By combining differentiable rendering with explicit point-based scene representations, 3D Gaussian Splatting (3DGS) has demonstrated breakthrough 3D reconstruction capabilities. However, to date 3DGS has had limited impact on robotics, where high-speed egomotion is pervasive: Egomotion introduces motion blur and leads to artifacts in existing frame-based 3DGS reconstruction methods. To address this challenge, we introduce Event3DGS, an event-based 3DGS framework. By exploiting the exceptional temporal resolution of event cameras, Event3GDS can reconstruct high-fidelity 3D structure and appearance under high-speed egomotion. Extensive experiments on multiple synthetic and real-world datasets demonstrate the superiority of Event3DGS compared with existing event-based dense 3D scene reconstruction frameworks; Event3DGS substantially improves reconstruction quality (+3dB) while reducing computational costs by 95%. Our framework also allows one to incorporate a few motion-blurred frame-based measurements into the reconstruction process to further improve appearance fidelity without loss of structural accuracy.} }
Endnote
%0 Conference Paper %T Event3DGS: Event-Based 3D Gaussian Splatting for High-Speed Robot Egomotion %A Tianyi Xiong %A Jiayi Wu %A Botao He %A Cornelia Fermuller %A Yiannis Aloimonos %A Heng Huang %A Christopher Metzler %B Proceedings of The 8th Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2025 %E Pulkit Agrawal %E Oliver Kroemer %E Wolfram Burgard %F pmlr-v270-xiong25b %I PMLR %P 4100--4118 %U https://proceedings.mlr.press/v270/xiong25b.html %V 270 %X By combining differentiable rendering with explicit point-based scene representations, 3D Gaussian Splatting (3DGS) has demonstrated breakthrough 3D reconstruction capabilities. However, to date 3DGS has had limited impact on robotics, where high-speed egomotion is pervasive: Egomotion introduces motion blur and leads to artifacts in existing frame-based 3DGS reconstruction methods. To address this challenge, we introduce Event3DGS, an event-based 3DGS framework. By exploiting the exceptional temporal resolution of event cameras, Event3GDS can reconstruct high-fidelity 3D structure and appearance under high-speed egomotion. Extensive experiments on multiple synthetic and real-world datasets demonstrate the superiority of Event3DGS compared with existing event-based dense 3D scene reconstruction frameworks; Event3DGS substantially improves reconstruction quality (+3dB) while reducing computational costs by 95%. Our framework also allows one to incorporate a few motion-blurred frame-based measurements into the reconstruction process to further improve appearance fidelity without loss of structural accuracy.
APA
Xiong, T., Wu, J., He, B., Fermuller, C., Aloimonos, Y., Huang, H. & Metzler, C.. (2025). Event3DGS: Event-Based 3D Gaussian Splatting for High-Speed Robot Egomotion. Proceedings of The 8th Conference on Robot Learning, in Proceedings of Machine Learning Research 270:4100-4118 Available from https://proceedings.mlr.press/v270/xiong25b.html.

Related Material