雷达科学与技术2026，Vol.24Issue(3)：299-307,9.DOI:10.3969/j.issn.1672-2337.2026.03.007
GS-CA:基于不确定性驱动的跨模态雷达相机三维目标检测
GS-CA:Uncertainty-Driven Cross-Modal 3D Object Detection with Radar and Camera
摘要
Abstract
This work investigates the advantages of millimeter-wave radar-camera fusion over vision-only ap-proaches and addresses the challenge of dynamic modality reliability.Therefore,a new multimodal fusion framework,GS-CA,an adaptive camera-radar fusion method for handling 3D perception tasks in the BEV perspective,is designed by introducing the discrete selection mechanism of the Gumbel-Softmax differentiable.The description of feature uncer-tainty is added during fusion and incorporated into the model training to increase the interpretability and provide im-provement directions for model optimization.Additionally,a corresponding deformable cross-attention mechanism is de-signed to increase the interaction of the two-modal BEV features.In the model evaluation,the methods show consider-able performance on the nuScenes dataset,and the target detection model demonstrates promising results in rainy and dark scenes in addition to ordinary scenes.Overall,the proposed framework establishes a new paradigm for multimodal fusion and offers valuable insights for the advancement of sensor fusion technologies.关键词
多模态融合/3D感知/特征不确定性/可变形交叉注意力/传感器融合Key words
multimodal fusion/3D perception/feature uncertainty/deformable cross-attention/sensor fusion分类
信息技术与安全科学引用本文复制引用
蒋留兵,刘仁超,车俐,吕连辉..GS-CA:基于不确定性驱动的跨模态雷达相机三维目标检测[J].雷达科学与技术,2026,24(3):299-307,9.基金项目
国家自然科学基金(61561010) （61561010）
广西创新驱动发展专项(桂科AA21077008) （桂科AA21077008）
广西无线宽带通信与信号处理重点实验室2022年主任基金(GXKL06220102,GXKL06220108) （GXKL06220102,GXKL06220108）
八桂学者专项经费资助(2019A51) （2019A51）
桂林电子科技大学研究生教育创新计划资助(2022YXW07,2022YCXS080) （2022YXW07,2022YCXS080）
2022年广西高等教育本科教学改革工程项目(2022JGB196) （2022JGB196）
桂林电子科技大学学位与研究生教改项目(2022YXW07,2023YXW02) （2022YXW07,2023YXW02）
广西研究生教育创新计划资源(YCSW2022271) （YCSW2022271）
