OpenREAD: Reinforced Open-Ended Reasoing for End-to-End Autonomous Driving with LLM-as-Critic Paper • 2512.01830 • Published Dec 1, 2025 • 3 • 2