On October 1, Beijing time, on Tesla 2022 AI Day, Tesla officially released its first mass-produced humanoid robot - Tesla Bot. It officially showed the robot's movements such as walking, squatting, and carrying items. Tesla Bot can even complete flowers and complete some relatively fine work in the factory. Officials said that after achieving large-scale mass production, the price of the robot is expected to drop to around 20,000 US dollars. In addition, Tesla's latest autonomous driving technology and the latest progress of Dojo supercomputers were also released on 2022 AI Day.
official introduction, Tesla robots use object boundaries recognized by all-visual means, and use the Tesla model's FSD supercomputing platform for training. Tesla Bot borrows the engineering development logic of Tesla models and uses the model's visual nerves and FSD computing platform to reduce the wiring and mechanical connections of the limbs. At the same time, the human bones are also simulated, referring to the pressure curve of each joint, and using mechanical structures to simulate. Let the robot have better pressure averages when moving the knees and other structures.
Tesla robot weighs 73kg, and a 2.3kWh battery pack is built into the torso to power the whole body, and the sensors and control systems are also concentrated in the torso. The central brain comes from Tesla car , and more wireless connection modules are added. The robot's engineers also simulate the robot's fall scene based on car impact tests to protect the central brain. After efficient integration, the robot consumes little power, with approximately 100w during sitting and 500w during slow travel.
official said that after achieving large-scale mass production, the price of the robot is expected to drop to around 20,000 US dollars. Musk said: "From now on, with the expansion of production scale and the decline in costs, the practicality of humanoid robots will increase year by year. In the future, a home robot may be cheaper than a car. Perhaps in less than ten years, people can buy a robot for their parents as a birthday gift."
was the first to propose the Tesla Bot concept at the 2021 AI Day held on August 20 last year, but the Tesla Bot concept that took place at that time was still played by humans. Just one year after the concept was proposed, Tesla will showcase the real Tesla Bot entity, and we have to sigh at Tesla's efficiency. Tesla CEO Musk proposed in an article published in China Netcom magazine not long ago: "The initial positioning of Tesla robots was to replace people in repetitive and boring and dangerous work. But the long-term goal is to serve thousands of households, such as cooking, mowing lawns, taking care of the elderly, etc.." What’s more interesting is that Musk immediately stated on social media that Tesla Bot will definitely launch the “Catwoman” version of the robot. Are you looking forward to it more?
In addition, Tesla has demonstrated the progress of FSD (full autonomous driving capability). Based on the previous basis, the recognition capability and execution decisions have been optimized. The FSD Beta version can already drive automatically from the parking lot to another parking lot, pass the intersection, and wait for the traffic light . Now FSD Beta can predict the surrounding environment and act, predict the next actions of other vehicles and pedestrians on the road, and give corresponding instructions to the vehicles.
In the unprotected left turn scenario, the latest FSD Beta will slow down first, choose a relatively safe route, be more anthropomorphic, and can even pass through the gap between vehicles and pedestrians. Previously, the FSD Beta would stop at the intersection at a relatively fast speed and wait for the vehicle to pass before passing. This upgraded trajectory judgment requires a lot of simulation calculation and simulation. Tesla uses the neural network learning to do a lot of work and analysis to ensure that the vehicle passes safer and more comfortable. FSD Beta's deep neural network reaches 1 million parameters, 150,000+ neural network layers, and 375,000 connections. Scenes that did not appear in reality can be simulated within 5 minutes.
Tesla is currently deploying the FSD beta version widely in North America and will launch large-scale testing of fully autonomous driving capability (FSD) software this year, with the number of users tested has expanded from 2,000 to 160,000 in 2021. The cumulative mileage of the beta version of the software has exceeded 40 million miles, and it is expected to exceed 100 million miles by the end of the year.
And the biggest news about Dojo supercomputer is that Musk also said that Dojo supercomputer will be provided to other users in the cloud in the future like the Amazon web service. The single training module of the Dojo supercomputer is made of a whole wafer, and each training module contains 25 Tesla's independently developed D1 neural network training chips. Since the entire wafer is not cut, each D1 chip is seamlessly connected together, and the latency between adjacent chips is extremely low. The training module realizes bandwidth retention to the greatest extent, and is combined with Tesla's own high-bandwidth and low-latency connectors; in a volume of less than 1 cubic foot, the computing power is as high as 9PFLOPs (90 trillion times) and the I/O bandwidth is as high as 36TB/s.
