Reproduce “Driver Assistant: Persuading Drivers to Adjust Secondary Tasks Using Large Language Models” 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
Wei Xiang, Muchen Li, Jie Yan, Manling Zheng, Hanfei Zhu, Mengyun Jiang, Lingyun Sun
Researchers at Zhejiang University developed an LLM-based persuasion system for Level 3 automated driving that proactively guides drivers to manage secondary tasks. Empirical results showed that the system led to significantly fewer secondary tasks, improved attention allocation, and reduced cognitive load compared to traditional alert systems in simulated driving scenarios.
Key insights
- Drivers prefer brief, relaxed, and interactive persuasion over commanding alerts, favoring direct advice that avoids emphasizing negative consequences.
- LLMs can effectively process real-time contextual information from driving environments and driver states to make nuanced decisions on when and how to persuade, demonstrating high consistency with human expert judgment.
- Proactive, humanized persuasion, as opposed to reactive warnings, is more effective in maintaining appropriate driver attention and reducing cognitive load during the dynamic challenges of Level 3 automation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.