Researchers at Delft University of Technology have developed a system that allows passengers to influence the driving style of autonomous vehicles using natural language commands. The system employs a large language model, specifically OpenAI's GPT-4o-mini, to interpret passenger requests like "I am running late, go fast" and translate them into adjustments for the vehicle's motion planner. This approach tunes the parameters of a safety-aware motion-planning algorithm rather than giving direct control to the LLM, ensuring the vehicle remains within safe operational bounds. The system also provides feedback to the passenger and requires confirmation before implementing changes, with simulations showing it successfully adjusted driving speed and smoothness according to user instructions. AI
IMPACT Enables more personalized and adaptable autonomous vehicle experiences by allowing natural language control of driving style.
RANK_REASON Research paper on using LLMs to control autonomous vehicle driving style. [lever_c_demoted from research: ic=1 ai=0.7]
- arXiv
- Delft University of Technology
- Diego Martinez-Baselga
- GPT-4o-mini
- IEEE Intelligent Transportation Systems Conference
- IEEE Journal Watch
- IEEE Spectrum
- IEEE Xplore
- Netherlands
- OpenAI
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
