| 10 | 0 | 80 |
| 下载次数 | 被引频次 | 阅读次数 |
针对无地图开放场景下交互风险感知粗糙、特征易过度平滑的问题，文中提出一种动态交互多模态轨迹预测模型。模型通过时空自适应精细化网格(STAR-Grid)引入碰撞时间倒数与最大值聚合，强化局部高危威胁感知；利用双向时空交叉注意力(Bi-SCA)编码器，解耦微观运动突变与宏观驾驶意图；结合拉普拉斯混合密度网络与赢者通吃(WTA)策略输出多模态轨迹。ETH/UCY数据集实验表明，模型平均位移和最终位移误差分别低至0.19 m和0.35 m,碰撞压低至1.28%。
Abstract：To address the issues of coarse perception of interaction risks and excessive feature smoothing in mapless open-world scenarios, this paper proposes a dynamic interactive multimodal trajectory prediction model. The model employs STAR-Grid to introduce the inverse of collision time and maximum value aggregation, thereby enhancing the detection of local high-risk threats; utilizes the Bi-SCA bidirectional spatiotemporal cross-attention encoder to decouple micro-level motion changes from macro-level driving intentions; and combines a Laplace mixture density network with a WTA strategy to generate multimodal trajectories. Experiments on the ETH/UCY datasets demonstrate that the model achieves average and final displacement errors as low as 0.19 m and 0.35 m, respectively, with a collision rate reduced to 1.28%.
[1] World H O.Global status report on road safety 2023[EB/OL].(2023-12-23)[2026-5-26].https://www.who.int/publications/i/item/9789240086517.
[2] 中华人民共和国公安部.全国机动车保有量达4.35 亿辆驾驶达5.23 亿人新能源汽车保有量超过2000万辆[EB/OL](2024-01-11)[2026-5-26].https://www.mps.gov.cn.
[3] Alahi A,Goel K,Ramanathan V,et al.Social LSTM:Human trajectory prediction in crowded spaces[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Las Vegas:IEEE,2016:961-971.
[4] Gupta A,Johnson J,Fei-Fei L,et al.Social GAN:Socially acceptable trajectories with generative adversarial networks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Salt Lake City:IEEE,2018:2255-2264.
[5] Hochreiter S,Schmidhuber J.Long short-term memory[J].Neural Computation,1997,9(8):1735-1780.
[6] Mohamed A,Qian K,Elhoseiny M,et al.Social-STGCNN:A social spatio-temporal graph convolutional neural network for human trajectory prediction[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Seattle:IEEE,2020:14424-14432.
[7] Wong C,Xia B,Zou Z,et al.SocialCircle:Learning the angle-based social interaction representation for pedestrian trajectory prediction[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Seattle:IEEE,2024:19005-19015.
[8] Liu Y,Hu T,Zhang H,et al.iTransformer:Inverted transformers are effective for time series forecasting[C]//International Conference on Learning Representations,2024:1-8.
[9] Gu T,Chen G,Li J,et al.Stochastic trajectory prediction via motion indeterminacy diffusion[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.New Orleans:IEEE,2022:17113-17122.
[10] 刘冰，张顺.基于多模态深度学习的源代码漏洞检测[J].长春工业大学学报，2025,46(3):193-199.
[11] 韩梅，胡黄水，王出航.基于多特征交叉融合的语音情感识别方法[J].长春工业大学学报，2025,46(3):216-221.
[12] Pellegrini S,Ess A,Schindler K,et al.You′ll never walk alone:Modeling social behavior for multi-target tracking[C]//IEEE 12th International Conference on Computer Vision.Kyoto:IEEE,2009:261-268.
[13] Lerner A,Chrysanthou Y,Lischinski D.Crowds by example[J].Computer Graphics Forum,2007,26(3):655-664.
基本信息:
DOI：10.15923/j.cnki.cn22-1382/t.2026.3.03
中图分类号:U463.6;TP18
引用信息:
[1]黄俊凯,崔高健,武广鑫,等.基于时空交叉注意力的无地图多模态轨迹预测[J].长春工业大学学报,2026,47(03):211-218.DOI:10.15923/j.cnki.cn22-1382/t.2026.3.03.
基金信息:
吉林省科技厅国际合作项目(20240402061GH)
2026-06-15
2026-06-15