Recent content by ChrML

For that steering wheel to work, there must be a speed dependent dynamic steering/turn ratio. So the steering wheel never turns that much.

These are solved technical problems though that won't take their programmers long to implement. Just let the car grab the public keys for associated apps from Tesla's servers with 7 days expiration every day, and it will know how to authenticate your phone without internet as long as car has...

Nah, this is no problem. The chassis is stiff and can handle it.

Simple. Like this: (my car)

As a software development leader, I would think FSD software on HW3 without self-driving enabled is a more efficient path for Tesla. Use the exact same software no matter what features you have, and just toggle them based on what the user has paid for. In this way all HW3 users will get the same...

Given a good architecture, the driving policy rules should be defined at such a high level that it should be testable without real life testing. Manual testing is extremely slow and unrepeatable, and will inhibit development speed a lot. The automated testing kit should be fully decoupled from...

Having the most advanced FSD system is actually a bad thing though. A simpler system is always better than more complex systems, providing it solves the same tasks. A better phrased question is, which FSD solves most problems and is most reliable?

Not necessarily related to time. FSD has been a separate branch for a long time, being independent of the released master branch. Occasionally they merge master into other branches like FSD to bring it up to date, and minimizing the amount of merge conflicts by resolving them along the way...

There are significant costs involved with training in datacentres such as Azure. Not on small scale, but Tesla has a huge dataset, big network and rapid/continuous iterations. Seems Tesla wants a reinforcing feedback loop where the training data/labeling maintains itself based on real world...

But it's a start, and a much welcome update. As long as there are good leaps at regular intervals, they're good for now.

The update we see now is hopefully the beginning of a ketchup effect. It usually happens in software when you do an architecture revamp that a lot of things down the road depend on. Until you eventually hit the local maximum of the new architecture. So far, so good.

Cases are usually added as they occur in production. Which means the test suite covers more edge scenarios the longer it's tested in production, regardless if you're on codebase 1, 2 or 3. Hopefully every Nth thousand test scenario reproduced in simulation eventually add another 9 to the...

This is true, however as you probably know 75% of modern software development is about testing frameworks, writing test scenarios and collecting/labeling test data. The actual code is only 25% of the work. Done correctly you can do fairly substantial rewrites, and validate it against your...

Really? The thread title says otherwise. They might succeed, or they might not, or they might deploy it to a few cars in a geofenced area to check of the "System deployed"- milestone for the investors. A date that far ahead suggests it's a goal, or an aspirational target date. It's pretty much...

Self-driving is still an arena where no products have really materialized on the market yet. Obviously because it's really hard (especially from 99% to 100%), but the rewards are insane. Just consider robo-taxis replacing every taxi driver and delivery truck in the world, Amazon automatically...