Tesla Optimus Update Technical Deep Dive (End-to-end neural network, teleoperation, force sensor)

Tesla just announced their new Tesla Optimus Update on May 2024. I'll be doing a deep dive on the technical details of this update and what it means. I will explain the end-to-end neural network, force sensor, tactile sensor, 2d camera, teleoperation, and more!
Thanks for watching! If you found this video helpful, please like, subscribe and share:
/ @kevinwoodrobotics
Social:
LinkedIn: / kevinwoodrobotics
Github: github.com/kevinwoodrobotics
Instagram: / kevinwoodrobotics
Twitter: / kevinwoodrobotics

[@kevinwoodrobotics]

FSD 12.3.6 vs 12.3.3 Same Route Comparison Side By Side
th-cam.com/video/QclOicosYPI/w-d-xo.html

[@obensustam3574]

It was a really informative and interesting video! 🤖😎

[@kevinwoodrobotics]

Thanks!

[@GETREALLYRICH]

Why not use wheels that can go up standard stairs? And a spine and arms that extend and contract?

[@kevinwoodrobotics]

That’s possible too. But I think they want to make it like humans for human made environments

[@GETREALLYRICH]

@@kevinwoodrobotics it would be more efficient for energy consumption most likely depending on the wheel design and faster moving across flat even surfaces , correct? Wheels could be placed in the foot and drop/up if the bit needs to move over uneven surfaces.

[@kevinwoodrobotics]

@@GETREALLYRICH yeah definitely more energy efficient with wheels but more versatile with legs

[@GETREALLYRICH]

@@kevinwoodrobotics good to hear. Would wheels that retract into the foot or on the side of the ankle /foot be difficult to implement? That way you have wheels and feet for more purposes and terrain

[@Astra2]

Funny how people are still betting against Tesla. It used to be that practical electric cars were impossible, then when that was achieved it became long-range electric cars. Oh, but they could never make an electric truck... When that was achieved people started saying FSD was impossible. Now, as FSD approaches full-scale roll-out people are saying Optimus will never be practical. Anyway, great breakdown.

[@kevinwoodrobotics]

Thanks! Yeah a lot of pessimism out there. Will be a question of when not if.

[@aravindsairam3432]

Still looks so fragile maybe they should switch to the what Boston dynamics is doing with their new Atlas bots. I guess eventually everyone will switch to their mechanics.

[@kevinwoodrobotics]

Yeah bostons new atlas looks pretty robust

[@kittehboiDJ]

I was thinking these guys need robot dogs.

[@jloiben12]

This is not impressive at all. This is far from being a leading edge product

[@kevinwoodrobotics]

Lots of work left

[@aahhhhhhhhhhhhh]

I feel like tesla is going for affordability rather than efficiency unfortunately. They want to sell as much as possible. Its pretty clear with how clean the robots design looks

[@DerekDavis213]

Optimus is putting batteries into a plastic case? Very slow. A human would be much faster.

[@kevinwoodrobotics]

Yeah lots of room for improvement

[@DerekDavis213]

@@kevinwoodrobotics How long do you think it will be, until Optimus can do these tasks for an average family?
grocery shopping
cook dinner
do the dishes
do the laundry
wash the car

[@kevinwoodrobotics]

@@DerekDavis213 very good question. I think for more controlled activities it’s probably already possible now. But to be reliable most of the time maybe another 1-2 years. They haven’t shown too much how they are training so hard to tell. But if they have enough training data it definitely possible. The speed of the task is still another issue though. So even if they can do it, might be too slow at first

[@DerekDavis213]

@@kevinwoodrobotics I think Tesla is greatly overstating the impact of 'training data'. If I was a betting man, I would bet BIG money that in 2 years Optimus cannot do that list of activities above, in a consumers home.

[@kevinwoodrobotics]

@@DerekDavis213 well if it’s Elon musks 1-2 years it’s probably actually 5-10 years 😂