融合时空查询的情景感知多模态车辆轨迹预测模型
Context-aware Multimodal Vehicle Trajectory Prediction Model Based on Spatio-Temporal Queries
摘要
Abstract
With the rapid development of autonomous driving technology,accurately predicting the movement trajectories of sur-rounding vehicles has become the key to ensuring driving safety.Most of the existing methods do not fully consider the interactions be-tween vehicles and the environment,as well as among vehicles and environmental scenario information,and have poor trajectory predic-tion performance in complex traffic scenarios.Based on this,a Context-aware Multimodal Vehicle Trajectory Pediction Model Based on Spatio-Temporal Query Transformer(STQformer)is proposed to efficiently understand and predict vehicle behaviors in complex traffic environments.The model is based on the Transformer framework,introduces learnable spatio-temporal queries and utilizes the social interaction module to achieve in-depth perception of vehicle intentions and more accurate trajectory prediction.The experimental results show that:Compared with the current advanced trajectory prediction algorithms,the performance of STQformer in long-term prediction has improved by 9%compared with the best-performing comparison model.This model is conducive to enhancing the safety and relia-bility of the autonomous driving system,promoting the development and application of autonomous driving technology,enabling it to better adapt to the complex and changeable traffic environment,reduce traffic accidents and improve traffic efficiency.关键词
车辆轨迹预测/时空查询/Transformer/多头注意力机制/深度学习Key words
vehicle trajectory prediction/spatio-temporal queries/Transformer/multi-head attention mechanism/deep learning分类
信息技术与安全科学引用本文复制引用
李庆,乔少杰,陈浩,李任杰,蒋宇河,李洲,刘晨旭,卓小军,韩楠..融合时空查询的情景感知多模态车辆轨迹预测模型[J].无线电通信技术,2025,51(3):454-465,12.基金项目
国家自然科学基金(62272066) （62272066）
四川省科技计划资助(2025ZNSFSC0044,2025YFHZ0194) （2025ZNSFSC0044,2025YFHZ0194）
成都市技术创新研发项目重点项目(2024-YF08-00029-GX) （2024-YF08-00029-GX）
成都市技术创新研发项目(2024-YF05-01217-SN) （2024-YF05-01217-SN）
成都市区域科技创新合作项目(2025-YF11-00050-HZ) （2025-YF11-00050-HZ）
网络空间安全教育部重点实验室及河南省网络空间态势感知重点实验室开放基金课题(KLCS20240106) （KLCS20240106）
网络空间大数据智能安全教育部重点实验室开放基金课题(CBDIS202404) National Natural Science Foundation of China(62272066) （CBDIS202404）
Sichuan Science and Technology Program(2025ZNSFSC0044,2025YFHZ0194) （2025ZNSFSC0044,2025YFHZ0194）
Chengdu Technological Innovation Research and Development Major Project(2024-YF08-00029-GX) （2024-YF08-00029-GX）
Chengdu Technological In-novation Research and Development Project(2024-YF05-01217-SN) （2024-YF05-01217-SN）
Chengdu Regional Science and Technology Innovation Cooperation Project(2025-YF11-00050-HZ) （2025-YF11-00050-HZ）
Open Foundation of Key Laboratory of Cyberspace Security,Ministry of Education of China and Henan Key Laboratory of Cy-berspace Situation Awareness(KLCS20240106) （KLCS20240106）
Open Research Fund of Key Laboratory of Cyberspace Big Data Intelligent Security(Chongqing Univer-sity of Posts and Telecommunications),Ministry of Education of China Under Grant(CBDIS202404) （Chongqing Univer-sity of Posts and Telecommunications）
