SCIENCE@home

TP-GLIF: Trajectory Prediction of Surrounding Vehicles in Unsignalized Roundabouts Based on Global-Local History Information Fusion

Yingnan Ye,Chunyan Wang,Wanzhong Zhao,Bo Zhang,Yujie Zhang

Published 2025 in IEEE transactions on intelligent transportation systems (Print)

ABSTRACT

For intelligent vehicles, accurate prediction of the trajectories of surrounding vehicles is an important basis for identifying potential risks in advance and making optimal decisions. However, unsignalized roundabouts have multiple entrance and exit ramps with uncertain angles and numbers, which leads to more diverse driving intentions and more complex multi-vehicle interactions of surrounding vehicles, which significantly reduces the accuracy of trajectory prediction and even causes wrong decisions and serious collision accidents. To this end, this paper innovatively proposes a method for predicting the trajectory of surrounding vehicles in unsignalized roundabouts based on global-local historical information fusion (TP-GLIF). From a global perspective, a scene-adaptive dynamic and static feature information encoder is constructed, and a multi-head attention mechanism is introduced to capture environmental information closely related to driving intention, so that the model always focuses on the dynamic and static features that are most beneficial to improving prediction accuracy. From a local perspective, a novel dynamic heterogeneous graph attention multi-vehicle interaction encoder is constructed, and a novel hierarchical Gaussian mixture model-hidden Markov model-support vector machine driving intention recognition algorithm is designed to guide prediction decoding. A long short-term decoder LS-GRU is designed to decode the surrounding vehicle trajectories under the influence of complex road structure features, complex multi-vehicle interaction features and diverse driving intention features in different time periods to reduce the cumulative error. Experimental results show that TP-GLIF can achieve accurate trajectory prediction in different roundabout scenes, and has good computational efficiency and generalization ability.

PUBLICATION RECORD

  • Publication year

    2025

  • Venue

    IEEE transactions on intelligent transportation systems (Print)

  • Publication date

    2025-12-01

  • Fields of study

    Computer Science, Engineering

  • Identifiers
    DOI 10.1109/TITS.2025.3605735
  • External record

    Open on Semantic Scholar

  • Source metadata

    Semantic Scholar

CITATION MAP

EXTRACTION MAP

CLAIMS

  • No claims are published for this paper.

CONCEPTS

  • No concepts are published for this paper.

REFERENCES

Showing 1-32 of 32 references · Page 1 of 1

CITED BY