From Observation to Prediction: LSTM for Vehicle Lane Change Forecasting on Highway On/Off-Ramps

2025-10-27

International Conference on Vehicular Electronics and Safety

AI Summary

This study addresses the challenge of predicting lane-changing behavior in highway on- and off-ramp zones, where complex vehicle interactions lead to low prediction accuracy and safety concerns. To tackle this issue, the authors propose a multi-layer LSTM model that performs temporal modeling of lane changes using the ExiD drone-based dataset. The work systematically compares prediction performance between ramp and regular straight-road scenarios. Experimental results demonstrate that, within a 4-second prediction horizon, the model achieves 76% accuracy in ramp areas and 94% on regular segments. This study provides the first empirical validation of the predictability of lane-changing behavior in highly interactive ramp environments, offering critical support for the development of intelligent transportation systems.

Technology Category

• Planning, Routing, and Scheduling: Model-Based Reasoning
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Social Networks and Social Media: Influence propagation, information diffusion, and the prediction on networks
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

On and off-ramps are understudied road sections even though they introduce a higher level of variation in highway interactions. Predicting vehicles’ behavior in these areas can decrease the impact of uncertainty and increase road safety. In this paper, the difference between this Area of Interest (AoI) and a straight highway section is studied. Multi-layered LSTM architecture to train the AoI model with ExiD drone dataset is utilized. In the process, different prediction horizons and different models’ workflow are tested. The results show great promise on horizons up to 4 seconds with prediction accuracy starting from about 76% for the AoI and 94% for the general highway scenarios on the maximum horizon.

Problem

• lane change forecasting
• on/off-ramps
• highway safety
• vehicle behavior prediction
• traffic uncertainty

Innovation

• LSTM
• lane change prediction
• on/off-ramps
• drone dataset
• highway interaction