Raphael Thys
A Tesla Optimus Robot showing off its Yoga skills.
Tesla Optimus, the humanoid robot from Tesla, has shown remarkable progress in its latest video posted by the official Tesla Optimus account. The video reveals that TeslaBot is now running on the same end-to-end neural network as the Tesla cars, which can process video input and generate control output.
The video begins with the Tesla Bot aka the Optimus robot performing a self-calibration routine, which is essential for adapting to new environments. It then shows how TeslaBot can use its vision and joint position sensors to accurately locate its limbs in space, without relying on any external feedback. This enables TeslaBot to interact with objects and perform tasks with precision and dexterity.
See Also
- Tesla Day: 'Optimus' AI robots may outnumber humans in future, claims Elon Musk
- Tesla's robot 'Optimus' arrives in a Tesla Store in New York
- Humanoid robot Apollo could rival Tesla's Optimus bot
Optimus can now sort objects autonomously 🤖Its neural network is trained fully end-to-end: video in, controls out.
Come join to help develop Optimus (& improve its yoga routine 🧘)
One of the tasks that Optimus demonstrates is sorting blue and green blocks into matching trays. Tesla Optimus can grasp each block with ease and sort them at a human-like speed. It can also handle dynamic changes in the environment, such as when a human intervenes and moves the blocks around. TeslaBot can quickly adjust to the new situation and resume its task. It can also correct its own errors, such as when a block lands on its side and needs to be rotated.
The video also showcases Tesla Bot’s balance and flexibility, as it performs some yoga poses that require standing on one leg and extending its limbs. These poses are not related to any practical workloads, but they show how TeslaBot can control its body and maintain its stability.