Tesla’s general-purpose robot uses advanced neurals & deep learning to do task such as sorting and yoga autonomously.
Tesla’s pet project Optimus, a “general purpose, bi-pedal, autonomous humanoid robot” was teased to sort objects by color without being trained to do so. More importantly, it was also displayed to do yoga, which is a remarkable feat for a humanoid robot as the weight distribution considerations are abnormally high for a robot.
Capable of self-calibrating and equipped with advanced neural networks, vision, and joint position encoders, the robot can accomplish general-purpose tasks such as sorting and can be trained to learn additional tasks, such as un-sorting.
The robot can use its vision to precisely locate its limbs in space, use vision to identify objects, and do tasks all with the help of deep learning.
In an announcement Tweet, the Tesla Optimus page said that the Optimus’ neural network is trained fully end-to-end.
Here is the video of Optimus performing the tasks:
Tesla’s AI division is hiring more experts.