Connecting The Dots
patreon
AUTONOMOUS DRIVING IN TESLA FACTORIES
🕑 Added 2025-01-29 10:55:04 +0000 UTCMembers/Patrons only
Tesla AI just posted the following video on X.
"Teslas now drive themselves from
their birthplace at the factory to their
designated loading dock lanes without
human intervention
One step closer to large-scale
unsupervised FSD"
https://x.com/Tesla_AI/status/1884457749226090590?t=WzvZGo_u72e-ZPz8JdwTFw&s=19
I mused about this years ago and SOOOOO waited for this to happen - but what Tesla showed is only Stage 1. Stay tuned - there is more to Come!
Doing factory routes is much easier than full autonomous driving on public roads. The environment is static, all routes are almost the same (so there is over fitting of training data), speed is limited and regulations on private property are more lenient
So deleting the need for a driver to take the car from factory to loading dock / parking lot is relatively easy, and saves costs on each car
These costs are driving time + return time in driver-return shuttle + some waiting for shuttle + percent of shuttle operation costs. Overall something like 15-20 minutes of driver employment cost
But that's just part of the story. The next stages are:
🔸 Automating first drive test - test drivers take cars on a short test route, to feel ride quality, suspension, steering feel, rattles+other NVH, etc. This can easily be done by utilizing the cars' existing sensors (microphone, accelerometers etc) while autonomously driving the test route.
What's more, results would be better and more consistent than with a human driver, and TBH I wouldn't be surprised if they're already doing that, or at least testing
🔸Automated loading from dock to Ro-Ro ship. Latest car-park navigation capabilities should make this doable. Will probably happen in a few months. Possibly a year bc I imagine that Ro-Ro operators would be conservative, but technically this too is definitely doable, and should save even more costs per car
🔸Automated loading and unloading on trucks. While this isn't part of regular training, Tesla can build a special training set with millions of such cases per year - passing the cars to a special loading mode (there's precedent in current factory mode), which should enable saving loading-driver costs as well as faster loading on trucks - which clears parking lots faster and enables larger factory throughput when lots are a limiting factor
Tesla's technologies keep on converging. We all know about their manufacturing lead, and people are starting to realize that besides being a product, Optimus will also reduce labor costs on making cars, trucks and energy packs. Now autonomous driving in factories will start to reduce costs as well💪
Can't help recalling this lyric... The future's so bright we gotta wear shades😎
Comments
J S
Great post! I just mentioned this on X, and I know this is a high - cotton and short-term issue, but if Tesla thinks solving FSD to be equal to or better than a human driver is imminent, why is it only $8K, why is it even available for sale, when it will be worth about $100K when this happens? Tesla previously ramped FSD up to $15K and then rapidly lowered it to $8K, but no mention of increasing the price. Do they really believe it isn't close? Are they just trying to increase their cash/profitability, collecting more $ at $8K vs less than half the takers at $15K? Is the current FSD only available for personal use, not able to be used as a robotaxi? Do we want a situation where speculators/investors arbitrage and get the difference between $8K and $100K?
Connecting The Dots
💯Using voice command - "Go charge yourself!"
Ulf Asplund
With charging going wireless, the full process at the factory goes automated. And eventually people who don't charge at home can send the car to get charged as they eat dinner.
