arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~

视觉与机器人

自动驾驶

自动驾驶感知、规划、BEV、占用预测、激光雷达和仿真评测。

2026-03-02 至 2026-03-02 共收录 4 信号源:cs.RO, cs.CV, eess.IV, cs.AI

1. BEV与占用 4 篇

2507.23599 2026-03-02 cs.CV 89%

DA-Occ: Direction-Aware 2D Convolution for Efficient and Geometry-Preserving 3D Occupancy Prediction in Autonomous Driving

DA-Occ:方向感知的2D卷积用于自动驾驶中高效且保持几何的3D占用预测

Yuchen Zhou, Yan Luo, Xiaogang Wang, Xingjian Gu, Mingzhou Lu, Xiangbo Shu

专题命中 BEV与占用 :autonomous driving(title,abstract);occupancy(title,abstract);BEV(abstract);分类 cs.CV

AI总结 DA-Occ通过方向感知卷积和高度分数投影,在自动驾驶中实现高效且保持几何的3D占用预测,平衡精度与效率。

Comments This manuscript has been withdrawn because we are preparing a significantly revised version with improved methodology and extended experimental evaluation

详情

展开后加载摘要…

URL PDF HTML 收藏
2602.23894 2026-03-02 cs.CV 83%

SelfOccFlow: Towards end-to-end self-supervised 3D Occupancy Flow prediction

SelfOccFlow:迈向端到端自监督3D占用流预测

Xavier Timoneda, Markus Herb, Fabian Duerr, Daniel Goehring

机构 * Onboard Fusion team at CARIAD SE, Volkswagen Group(CARIAD SE 车辆集团 onboard 融合团队) Dahlem Center for Machine Learning and Robotics group at Freie Universität Berlin(柏林自由大学 Dahlem 机器学习与机器人学小组)

专题命中 BEV与占用 :occupancy(title,abstract);autonomous driving(abstract);分类 cs.CV

AI总结 SelfOccFlow提出了一种端到端自监督方法,通过分解场景为静态和动态符号距离场并利用特征余弦相似性,实现无需人工标注的3D占用流估计。

Comments Accepted version. Final version is published in IEEE Robotics and Automation Letters, DOI: 10.1109/LRA.2026.3665447

Journal ref IEEE Robotics and Automation Letters, vol. 11, no. 4, pp. 4331-4338, 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2602.23575 2026-03-02 cs.CV cs.AI 73%

CycleBEV: Regularizing View Transformation Networks via View Cycle Consistency for Bird's-Eye-View Semantic Segmentation

CycleBEV: 通过视图循环一致性正则化视图转换网络用于鸟瞰图语义分割

Jeongbin Hong, Dooseop Choi, Taeg-Hyun An, Kyounghwan An, Kyoung-Wook Min

机构 * Electronics and Telecommunications Research Institute (ETRI)(电子电信研究院) University of Science and Technology (UST)(科学技术大学)

专题命中 BEV与占用 :autonomous driving(abstract);BEV(abstract);分类 cs.CV、cs.AI

AI总结 CycleBEV通过引入视图循环一致性正则化方法,提升视图转换网络在鸟瞰图语义分割中的性能,实现显著的mIoU提升。

Comments CVPR 2026

详情

展开后加载摘要…

URL PDF HTML 收藏
2602.23691 2026-03-02 cond-mat.supr-con 50%

Inverse Isotope Effect in the Ternary Perovskite Hydride SrPdH/D$_{2.9}$: A Signature of Quantum Zero-Point Fluctuations

反同位素效应在三元钙钛矿氢化物 SrPdH/D$_{2.9}$ 中:量子零点涨落的特征

Wencheng Lu, Mihir Sahoo, Roman Lucrezi, Michael J. Hutcheon, Shubham Sinha, Pedro N. Ferreira, Chris J. Pickard, Qiang Zhang, Matthew N. Julian, Rohit P. Prasankumar, Christoph Heil, Timothy A. Strobel

专题命中 BEV与占用 :occupancy(abstract)

AI总结 该研究通过第一性原理计算揭示了三元钙钛矿氢化物 SrPdH/D$_{2.9}$ 的超导性,并发现反同位素效应源于量子零点运动。

详情

展开后加载摘要…

URL PDF HTML 收藏