Prediction of Lane Change Intentions of Human Drivers using an LSTM, a CNN and a Transformer

2025-07-11

AI Summary

This study addresses the problem of predicting human drivers’ lane-changing intentions in complex traffic scenarios. We propose a sequence-based prediction framework that models intention over temporal intervals—rather than at isolated time points—to enhance interpretability and reliability. Leveraging the highD public dataset, we systematically evaluate and compare three deep learning architectures—LSTM, CNN, and Transformer—under varying input feature configurations, with emphasis on generalization capability and robustness against overfitting. Results demonstrate that the Transformer architecture consistently outperforms the others, achieving state-of-the-art accuracy (96.73%), precision, and recall, alongside superior generalization. Crucially, input design significantly influences performance, with temporally structured trajectory features proving most effective. This work contributes the first empirical, multi-architecture comparison for interval-based driver intention prediction and establishes a high-fidelity, interpretable framework for motion planning in autonomous driving—bridging critical gaps in both interval modeling and architectural benchmarking.

Technology Category

• Machine Learning: Deep Neural Architectures and Foundation Models
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Lane changes of preceding vehicles have a great impact on the motion planning of automated vehicles especially in complex traffic situations. Predicting them would benefit the public in terms of safety and efficiency. While many research efforts have been made in this direction, few concentrated on predicting maneuvers within a set time interval compared to predicting at a set prediction time. In addition, there exist a lack of comparisons between different architectures to try to determine the best performing one and to assess how to correctly choose the input for such models. In this paper the structure of an LSTM, a CNN and a Transformer network are described and implemented to predict the intention of human drivers to perform a lane change. We show how the data was prepared starting from a publicly available dataset (highD), which features were used, how the networks were designed and finally we compare the results of the three networks with different configurations of input data. We found that transformer networks performed better than the other networks and was less affected by overfitting. The accuracy of the method spanned from $82.79%$ to $96.73%$ for different input configurations and showed overall good performances considering also precision and recall.

Problem

• Predicting human driver lane change intentions for safety
• Comparing LSTM, CNN, Transformer performance in prediction
• Assessing input data impact on model accuracy

Innovation

• Uses LSTM, CNN, Transformer for lane change prediction
• Compares three networks with different input configurations
• Transformer outperforms others with less overfitting