The Tesla AI team’s “data engine”

Here’s a way to think of the potential usefulness of Tesla’s HW2 fleet. The development process is a feedback loop where Autopilot driving errors (flagged by disengagements, aborts, and crashes) lead engineers to identify a problem. That leads the engineers to design a trigger to collect sensor data relevant to that problem.

A slide from a talk by Andrej Karpathy, Tesla’s Director of AI:

With Tesla’s approach, you can actually test lane keeping for 600 million miles, catch errors, and try to fix the failure modes that cause those errors. Then push your fix to ~150,000 cars, watch how the fix works for 100 million miles (driven in under 3 months), and iterate.

1 Like