Road User

Understanding road user behavior is crucial for improving road safety and developing autonomous vehicles. Current research focuses on predicting user actions (e.g., pedestrian crossings, lane changes) using various methods, including deep reinforcement learning combined with cognitive models to simulate human decision-making under perceptual limitations, and knowledge graph-based approaches for explainable predictions. These efforts aim to create more accurate and human-like models, moving beyond simple error metrics to assess behavioral realism and incorporating infrastructure analysis to understand its influence on user actions. This work has significant implications for both improving traffic safety and advancing the development of reliable autonomous driving systems.