[ICRA 2024] SuperFusion: 多级别激光雷达-摄像头融合技术,用于远程高清地图生成
This repository contains the implementation of the ICRA 2024 paper:
SuperFusion: Multilevel LiDAR-Camera Fusion for Long-Range HD Map Generation
Hao Dong, Xianjing Zhang, Jintao Xu, Rui Ai, Weihao Gu, Huimin Lu, Juho Kannala and Xieyuanli Chen
Link to the arXiv version of the paper is available.
Update: We have a new survey paper on Multimodal Adaptation and Generalization
Pipeline overview of SuperFusion. Our method fuses camera and LiDAR data in three levels: the data-level fusion fuses depth information from LiDAR to improve the accuracy of image depth estimation, the feature-level fusion uses cross-attention for long-range LiDAR BEV feature prediction with the guidance of image features, and the BEV-level fusion aligns two branches to generate high-quality fused BEV features. Finally, the fused BEV features can support different heads, including semantic segmentation, instance embedding, and direction prediction, and finally post-processed to generate the HD map prediction.
High-definition (HD) semantic map generation of the environment is an essential component of autonomous driving. Existing methods have achieved good performance in this task by fusing different sensor modalities, such as LiDAR and camera. However, current works are based on raw data or network feature-level fusion and only consider short-range HD map generation, limiting their deployment to realistic autonomous driving applications. In this paper, we focus on the task of building the HD maps in both short ranges, i.e., within 30 m, and also predicting long-range HD maps up to 90 m, which is required by downstream path planning and control tasks to improve the smoothness and safety of autonomous driving. To this end, we propose a novel network named SuperFusion, exploiting the fusion of LiDAR and camera data at multiple levels. We use LiDAR depth to improve image depth estimation and use image features to guide long-range LiDAR feature prediction. We benchmark our SuperFusion on the nuScenes dataset and a self-recorded dataset and show that it outperforms the state-of-the-art baseline methods with large margins on all intervals. Additionally, we apply the generated HD map to a downstream path planning task, demonstrating that the long-range HD maps predicted by our method can lead to better path planning for autonomous vehicles.
Download nuScenes dataset (Full dataset and Map expansion) and unzip files. The folder should be like
Install dependencies by running
pip install -r requirement.txtDownload the pretrained DeepLabV3 model and place within the checkpoints directory
Download the pretrained SuperFusion model and place within the runs directory
python train.py --instance_seg --direction_pred --depth_sup --dataroot /path/to/nuScenes/ --pretrained --add_depth_channelpython evaluate_iou_split.py --dataroot /path/to/nuScenes/ --modelf runs/model.pt --instance_seg --direction_pred --depth_sup --add_depth_channel --pretrainedpython export_pred_to_json.py --dataroot /path/to/nuScenes/ --modelf runs/model.pt --depth_sup --add_depth_channel --pretrainedpython evaluate_json_split.py --result_path output.json --dataroot /path/to/nuScenes/python vis_prediction_gt.py --instance_seg --direction_pred --dataroot /path/to/nuScenes/python vis_prediction.py --modelf runs/model.pt --instance_seg --direction_pred --depth_sup --pretrained --add_depth_channel --version v1.0-trainval --dataroot /path/to/nuScenes/If you use our implementation in your academic work, please cite the corresponding paper:
@article{dong2022SuperFusion,
author = {Hao Dong and Xianjing Zhang and Jintao Xu and Rui Ai and Weihao Gu and Huimin Lu and Juho Kannala and Xieyuanli Chen},
title = {{SuperFusion: Multilevel LiDAR-Camera Fusion for Long-Range HD Map Generation}},
journal = {arXiv preprint arXiv:2211.15656},
year = {2022},
}SimMMDG: SimMMDG: A Simple and Effective Framework for Multi-modal Domain Generalization
MOOSA: Towards Multimodal Open-Set Domain Generalization and Adaptation through Self-supervision
AEO: Towards Robust Multimodal Open-set Test-time Adaptation via Adaptive Entropy-aware Optimization
Survey: Advances in Multimodal Adaptation and Generalization: From Traditional Approaches to Foundation Models
MultiOOD: Scaling Out-of-Distribution Detection for Multiple Modalities
Many thanks to the excellent open-source projects HDMapNet,LSS,AlignSeg and CaDDN.
暂无开放 Issues,或尚未同步最近议题。