AI for self-driving cars can now be tested for real-world failures, not just perfect conditions

Researchers found a new way to test the AI that helps self-driving cars stay in their lane. This method creates fake but realistic problems, like fog or sensor glitches, to see how the AI performs.

Companies building self-driving cars have relied on testing AI with clean data. They assumed it would work in the real world. This paper shows that assumption is wrong. It reveals huge performance drops under common conditions like fog. Developers can now find critical flaws in their AI systems before they are deployed on actual roads.