Real-Time Driver Intention Prediction for Lane Merging using Deep Neural Networks

This paper introduces a real-time framework for predicting driver intentions during lane merging using deep neural networks (DNNs). The main goal is to improve road safety and optimize traffic flow by accurately predicting a driver's intention to merge. This approach examines essential data inputs, including vehicle velocity, acceleration, surrounding traffic dynamics, and driver behavioral indicators. Trained the DNN model on a comprehensive dataset that includes diverse driving scenarios and conditions. Through extensive testing, the model achieved notable enhancements in predictive accuracy, attaining a 98.96% accuracy rate in reliable intention detection compared to traditional methods. The results indicate that utilizing advanced deep learning (DL) methodologies may significantly enhance the advancement of intelligent driver-assistance systems (ADAS). It highlights the significance of real-time processing capabilities in automotive applications, enabling improved safety measures and informed decision-making while driving. Future work will focus on verifying the concept in real-world scenarios.