ScalableMap: Scalable Map Learning for Online Long-Range Vectorized HD Map Construction

Jingyi Yu, Zizhao Zhang, Shengfu Xia, Jizhang Sang
Proceedings of The 7th Conference on Robot Learning, PMLR 229:2429-2443, 2023.

Abstract

We propose a novel end-to-end pipeline for online long-range vectorized high-definition (HD) map construction using on-board camera sensors. The vectorized representation of HD maps, employing polylines and polygons to represent map elements, is widely used by downstream tasks. However, previous schemes designed with reference to dynamic object detection overlook the structural constraints within linear map elements, resulting in performance degradation in long-range scenarios. In this paper, we exploit the properties of map elements to improve the performance of map construction. We extract more accurate bird’s eye view (BEV) features guided by their linear structure, and then propose a hierarchical sparse map representation to further leverage the scalability of vectorized map elements, and design a progressive decoding mechanism and a supervision strategy based on this representation. Our approach, ScalableMap, demonstrates superior performance on the nuScenes dataset, especially in long-range scenarios, surpassing previous state-of-the-art model by 6.5 mAP while achieving 18.3 FPS.

Cite this Paper

BibTeX
@InProceedings{pmlr-v229-yu23b, title = {ScalableMap: Scalable Map Learning for Online Long-Range Vectorized HD Map Construction}, author = {Yu, Jingyi and Zhang, Zizhao and Xia, Shengfu and Sang, Jizhang}, booktitle = {Proceedings of The 7th Conference on Robot Learning}, pages = {2429--2443}, year = {2023}, editor = {Tan, Jie and Toussaint, Marc and Darvish, Kourosh}, volume = {229}, series = {Proceedings of Machine Learning Research}, month = {06--09 Nov}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v229/yu23b/yu23b.pdf}, url = {https://proceedings.mlr.press/v229/yu23b.html}, abstract = {We propose a novel end-to-end pipeline for online long-range vectorized high-definition (HD) map construction using on-board camera sensors. The vectorized representation of HD maps, employing polylines and polygons to represent map elements, is widely used by downstream tasks. However, previous schemes designed with reference to dynamic object detection overlook the structural constraints within linear map elements, resulting in performance degradation in long-range scenarios. In this paper, we exploit the properties of map elements to improve the performance of map construction. We extract more accurate bird’s eye view (BEV) features guided by their linear structure, and then propose a hierarchical sparse map representation to further leverage the scalability of vectorized map elements, and design a progressive decoding mechanism and a supervision strategy based on this representation. Our approach, ScalableMap, demonstrates superior performance on the nuScenes dataset, especially in long-range scenarios, surpassing previous state-of-the-art model by 6.5 mAP while achieving 18.3 FPS.} }
Endnote
%0 Conference Paper %T ScalableMap: Scalable Map Learning for Online Long-Range Vectorized HD Map Construction %A Jingyi Yu %A Zizhao Zhang %A Shengfu Xia %A Jizhang Sang %B Proceedings of The 7th Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2023 %E Jie Tan %E Marc Toussaint %E Kourosh Darvish %F pmlr-v229-yu23b %I PMLR %P 2429--2443 %U https://proceedings.mlr.press/v229/yu23b.html %V 229 %X We propose a novel end-to-end pipeline for online long-range vectorized high-definition (HD) map construction using on-board camera sensors. The vectorized representation of HD maps, employing polylines and polygons to represent map elements, is widely used by downstream tasks. However, previous schemes designed with reference to dynamic object detection overlook the structural constraints within linear map elements, resulting in performance degradation in long-range scenarios. In this paper, we exploit the properties of map elements to improve the performance of map construction. We extract more accurate bird’s eye view (BEV) features guided by their linear structure, and then propose a hierarchical sparse map representation to further leverage the scalability of vectorized map elements, and design a progressive decoding mechanism and a supervision strategy based on this representation. Our approach, ScalableMap, demonstrates superior performance on the nuScenes dataset, especially in long-range scenarios, surpassing previous state-of-the-art model by 6.5 mAP while achieving 18.3 FPS.
APA
Yu, J., Zhang, Z., Xia, S. & Sang, J.. (2023). ScalableMap: Scalable Map Learning for Online Long-Range Vectorized HD Map Construction. Proceedings of The 7th Conference on Robot Learning, in Proceedings of Machine Learning Research 229:2429-2443 Available from https://proceedings.mlr.press/v229/yu23b.html.

Related Material