Toward zero traffic accidents: How we built a video AI agent for automated driving @WeightsBiases
Toward zero traffic accidents: How we built a video AI agent for automated driving  @WeightsBiases
Uploaded July 2025 | Updated September 2026, 2 weeks ago
To build safer automated driving systems, we test our software by driving thousands of hours of real-world on-road vehicle testing. Manually checking these driving videos was slow and costly, so we developed AutoTriage—a video AI agent that finds system errors and identifies their root causes. We'll share three key takeaways behind our success: creating high-quality datasets, working closely with domain experts, and quickly adopting advances in generative AI. You'll learn practical steps to take your AI agent from prototype to production.

Chapters:
0:00 – Introduction: Woven by Toyota & Mission
1:41 – Why Automated Driving Matters
3:50 – Toyota’s Approach to Preventing Accidents
5:01 – Understanding Driving: Perception, Planning, Control
6:01 – Identifying Bottlenecks in AV Development
7:49 – Launching AUTO: Using GenAI for Bug Triage
9:18 – Challenges Building a Video AI Agent
12:21 – Overcoming Failures with Data, Prompts, and Models
14:26 – From Single Agent to Multi-Agent Judging
15:22 – Key Takeaways: Data, Collaboration, Persistence, Experiments
17:01 – Final Vision: Safer Roads Through AI-Driven Automation
Toward zero traffic accidents: How we built a video AI agent for automated drivingThe AI that solves the market: A new era in forecasting with natural language explainability
Weights & Biases |

Toward zero traffic accidents: How we built a video AI agent for automated driving

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER