Reproduce “A Study on Learning and Simulating Personalized Car-Following Driving Style” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Shili Sheng, Erfan Pakdamanian, Kyungtae Han, Ziran Wang, Lu Feng
This study develops a robust method for learning and simulating personalized car-following driving styles, particularly for drivers with limited observational data. It demonstrates that personalized styles can be accurately predicted for new drivers with as few as six car-following events, and these styles can be effectively integrated into a personalized adaptive cruise control system that maintains safety.
Key insights
- Human driving preferences, specifically car-following styles, can be effectively represented as quantitative reward functions inferred through model-free Inverse Reinforcement Learning.
- Personalized driving styles for new drivers can be accurately predicted with limited observational data by leveraging similarities to pre-identified driving style clusters.
- Integrating learned personalized reward functions into a safety-constrained control framework enables the design of user-centric automated driving systems that balance individual comfort with safety.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.