Yunhao Bu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Yunhao Bu has a total of 8 co-authors. Here are the top co-authors by number of shared publications:
- Jianhang Liu
Qingdao Institute of Software, College of Computer Science and Technology, China University of Petroleum (East China), Qingdao, China
2 shared publications - Shanchen Pang
China University of Petroleum (East China)
2 shared publications Haibo Wu
Chinese Academy of Sciences
1 shared publication- Jiangwan Wu
Qingdao Institute of Software, College of Computer Science and Technology, China University of Petroleum (East China), Qingdao, China
1 shared publication - Mu Zhou
Chongqing University of Posts and Telecommunications
1 shared publication - Shibao Li
College of Oceanography and Space Informatics, China University of Petroleum (East China), Qingdao, China
1 shared publication
Latest Publications
- 1 October 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 75 (10) , 22216-22227
CoGFormer: A Context-Gated Spatial Transformer With Temporal-Spatial Gated Graph Convolution for Traffic Flow Forecasting
- 1 October 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 75 (10) , 22216-22227
CoGFormer: A Context-Gated Spatial Transformer With Temporal-Spatial Gated Graph Convolution for Traffic Flow Forecasting
- 11 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. PP (00189545) , 1-14
Efficient Hybrid Collaborative Perception Based on Ternary Spatial Confidence Map Communication
Multi-agent collaborative perception technology is attracting considerable attention because it can extend the perception range of autonomous vehicles, and thus improve road safety. However, challenges such as communication redundancy, low perception precision, and a single collaboration mode adversely impact the accuracy and timeliness of environmental perception, especially under bandwidth and latency constraints in complex real-world traffic scenarios. To address the above problems, we design a hybrid collaborative sensing architecture guided by a ternary spatial confidence map. First, we propose a hybrid communication strategy based on the discretization of the classification confidence map and regional overlap assessment, which effectively mitigates communication redundancy among agents by dynamically guiding the selective transmission of intermediate features and late detection results. Furthermore, a dual-branch intermediate feature fusion method is designed, together with a collaborative supplementation mechanism for late detection results. This method combines convolutions with transformers to capture local structures and model global semantics, thereby improving perception accuracy in real-world scenarios. Comprehensive experiments on the OPV2V and DAIR-V2X-C datasets demonstrate that the proposed method effectively reduces communication redundancy and significantly improves perception accuracy in practical environments. Specifically, on the DAIR-V2X-C dataset, compared to existing methods, the proposed method reduces the accuracy loss by approximately 10% at AP@0.5.
- 11 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. PP (00189545) , 1-14
Efficient Hybrid Collaborative Perception Based on Ternary Spatial Confidence Map Communication
Multi-agent collaborative perception technology is attracting considerable attention because it can extend the perception range of autonomous vehicles, and thus improve road safety. However, challenges such as communication redundancy, low perception precision, and a single collaboration mode adversely impact the accuracy and timeliness of environmental perception, especially under bandwidth and latency constraints in complex real-world traffic scenarios. To address the above problems, we design a hybrid collaborative sensing architecture guided by a ternary spatial confidence map. First, we propose a hybrid communication strategy based on the discretization of the classification confidence map and regional overlap assessment, which effectively mitigates communication redundancy among agents by dynamically guiding the selective transmission of intermediate features and late detection results. Furthermore, a dual-branch intermediate feature fusion method is designed, together with a collaborative supplementation mechanism for late detection results. This method combines convolutions with transformers to capture local structures and model global semantics, thereby improving perception accuracy in real-world scenarios. Comprehensive experiments on the OPV2V and DAIR-V2X-C datasets demonstrate that the proposed method effectively reduces communication redundancy and significantly improves perception accuracy in practical environments. Specifically, on the DAIR-V2X-C dataset, compared to existing methods, the proposed method reduces the accuracy loss by approximately 10% at AP@0.5.