Exclusive: Waymo says there's no AI shortcut to self-driving
Summary
Waymo says there is no quick shortcut to creating fully safe self-driving cars using AI alone. After over 15 years and 200 million miles driven by its autonomous vehicles, Waymo emphasizes combining many tools like cameras, lidar, radar, and careful testing to ensure safety.Key Facts
- Waymo has been developing self-driving cars since 2009 and has logged over 200 million driverless miles.
- The company uses multiple specialized AI models for tasks like detecting pedestrians and traffic lights.
- Waymo argues that relying on only one large AI system ("end-to-end" neural networks) is not safe enough.
- Competitors like Tesla focus on camera-only systems with end-to-end AI, which Waymo says may miss important details.
- Waymo believes combining different sensors (cameras, lidar, radar) gives better visibility for AI to work safely.
- The company stresses that fixing AI mistakes in the physical world is harder than in software because you cannot just reboot a car.
- Waymo published a blog to share 10 lessons from its experience and to encourage safety standards in the industry.
- Despite advances in AI, Waymo maintains there is no single solution to quickly achieve safe autonomous driving.
Read the Full Article
This is a fact-based summary from The Actual News. Click below to read the complete story directly from the original source.
