Vehicle Trajectory: Intelligent Vehicle Behavior Intention Recognition Based on Catboost

This paper explores the technology of intelligent vehicle behavior intention recognition based on the CatBoost model. The technological framework of intelligent vehicles includes three core modules: environmental perception, decision-making and planning, and motion control. Directly inputting perception information into the decision-making module faces challenges in complex and dynamic urban traffic environments, which may lead to suboptimal solutions and trajectories that are inconsistent with driving habits or violate traffic regulations. To enhance the safety, comfort, and operational efficiency of intelligent vehicles, accurately predicting the future behaviors and trajectories of surrounding traffic participants has become a focal point of research. The CatBoost model has shown significant advantages in the field of intelligent transportation due to its inherent ability to handle categorical features, excellent generalization capabilities, and high accuracy. This study validates the effectiveness of the CatBoost model in vehicle behavior recognition using the HighD dataset and compares it with other models such as LightGBM and XGBoost. The results indicate that CatBoost performs better in terms of behavioral classification accuracy and low false positive rate, providing reliable support for the decision-making planning and safe operation of intelligent automotive systems.