Tesla humanoid robot Optimus unveiled FSD latest information on fully autonomous driving

Bitauto News On October 1, Beijing time, at Tesla’s 2022 AI Day to “flex its muscles” for recruitment, the humanoid robot Optimus was officially unveiled. Tesla says its price can be reduced to less than $20,000, making it cheaper than a car. In addition, Tesla also released the latest information on FSD fully autonomous driving and the Dojo supercomputer.

The humanoid robot Optimus is unveiled

The Optimus robot has 28 degrees of freedom throughout its body; its ergonomic hand design has 11 degrees of freedom; and it has a bionic joint design. The overall research and development is based on vehicle design, and safety is created based on automotive safety simulation analysis capabilities. The actuator can lift a piano weighing half a ton.

The humanoid robot Optimus will mainly handle physical work and a large amount of repetitive labor, allowing people to engage in more mental work. Musk said at the scene: "The robot we are showing now is the robot we are about to put into use." Tesla is confident that it can produce millions of humanoid robots in the future, and the price can be reduced to less than $20,000, making it cheaper than cars.

Compared with the robot concept released last year, Optimus will be heavier, weighing 73 kilograms, and equipped with the same FSD system as Tesla cars. Tesla has applied the training of cars to the robot. Through sensors and computer vision, it uses massive data to continuously train and become more and more powerful. Not only can it take on repetitive tasks, but it can also be your friend by your side.

The latest information on FSD fully autonomous driving technology

Tesla also announced the latest progress of FSD. The number of users participating in the FSD Beta version test has expanded from 2,000 in 2021 to 16 Thousands of people.

In the latest FSD version, Tesla will improve the decision-making optimization of intersection turns, and improve the safety and comfort of decision-making through the research and simulation of the trajectories of pedestrians and other vehicles in more complex scenarios. performance and make decisions within 100 milliseconds.

FSD Beta’s deep neural network reaches 1 million parameters, 150,00 neural network layers, and 375,000 connections. Scenarios that do not appear in reality can be simulated within 5 minutes. In the future, FSD will be more intelligent, continuously improve driving safety, and make every trip easier and smarter.

Dojo supercomputer

To achieve ultra-high computing power for artificial intelligence training, while expanding bandwidth, reducing latency, and saving costs, Tesla built the Dojo supercomputer. A single training module of the Dojo supercomputer consists of 25 D1 chips, a neural network training chip independently developed by Tesla. Tesla will launch the Dojo cabinet in the first quarter of 2023.

Each Dojo ExaPod integrates 120 training modules, has 3,000 built-in D1 chips, has more than 1 million training nodes, and has a computing power of 1.1 EFLOP (petaflops).

According to the "Hot List" data of Bitauto App, Tesla's average daily attention is 27,000, ranking 154th among all brands. If you need more data, please go to Bitauto App to view .