The Tesla Bot, Dojo supercomputer and more: Here’s what you should not miss about Tesla AI Day 2022

Tesla AI Day
Image credits: Tesla

Tesla Motors has long been known for its unique approach to electric vehicles and has pioneered self-driving, the latest in driving technology. 

On September 30, 2022, Tesla hosted an event — AI Day to give the public a better understanding of its current and future plans.

The event, which is focused on the company’s plans and progress in the realm of artificial intelligence, detailed its pursuits towards the Telsa bot, Full Self Driving, and Dojo supercomputer. While attendees received the event well, some critics raised concerns about the company’s ability to execute its ambitious plans. 

Elon Musk unveils ‘Optimus’ robot

In a bid to diversify its portfolio besides automobiles, Elon Musk, Tesla CEO, unveiled the much anticipated humanoid robot – Optimus. It’s worth mentioning here that the Optimus bot was first teased at last year’s AI Day. According to Musk, Optimus weighs around 72 kgs (160 pounds) and walks like a human. 

Tesla CEO also adds that Optimus would be equipped with a 2.3 kWh battery pack. The humanoid robot consumes 500 watts when walking and around 100 watts while sitting down. Further, Optimus will be both Wi-Fi and LTE capable and designed to be updated over time through OTA.

Post the event, Musk tweeted, “Naturally, there will be a catgirl version of our Optimus robot.”

Tesla intends to use Optimus in their factories to help with production and plans to mass produce the bot, with each costing around $20,000. However, many critics were not impressed with the unveiling of Optimus and said it was underwhelming. 

However, Dennis Hong, UCLA robotics expert, tweeted, “What was most impressive to me was what the Optimus team could accomplish in such a short period. If you are in this field, you would agree, too. The prototype they have created will be an excellent beginning platform for them to learn from and build upon.”

Hong tweeted about the mass production, “However, I disagree with their decision to mass produce the model shown at the event. I’m sure there are many good reasons behind it that I am not aware of, but I do not believe that is the one that can be used in a real world setting in any meaningful capacity.”

Full Self-driving grows to 1,60,000 beta users

Alongside Optimus, Tesla also provided updates on its Fully Self Driving (FSD) Beta. Tesla is now beta testing its full self-driving system with 1,60,000 vehicles. To put things in perspective, this number was around 2,000 last year. According to Tesla, FSD has had 35 software releases to date. According to Musk, the FSD tech will be ready for a global rollout by the end of this year. 

Tesla’s engineers have explained how they’ve managed to cut down the car’s decision-making time from milliseconds to just 100 microseconds – 10 times faster. The team showed how FSD (Full Self-Driving) technology works by mapping the world around Teslas in 3D geometry and basing decisions on what it sees.

Well, this doesn’t end here! Training the FSD model to make choices isn’t an easy task. As a result, Tesla put together three supercomputers, which are currently made up of 14,000 GPUs — 10,000 for training and 4,000 for labelling. 

Besides Tesla, there are other startups too, working on the self-driving car.

Dojo supercomputer

Thirdly, Tesla has unveiled its new Dojo supercomputer, which it says is the fastest and most powerful computer for training artificial intelligence (AI) models. The computer is designed for use in self-driving cars.

Tesla’s Dojo supercomputer is explicitly designed for AI machine learning and video training. It uses data from Tesla’s fleet of vehicles to train models that can be used for various purposes.

Dojo started as a chip and tile but has now become a system tray and an entire cabinet. Once completed, Dojo will be one of the most powerful supercomputers in the world, claims Tesla. 

According to Tesla, Dojo’s cabinets will consist of 120 tiles that will hold 3,000 custom D1 chips. In addition, each cabinet contains trays comprised of 2 x 3 tiles. As a result, each cabinet can produce 100 PFlops.

Upon initial tests, the supercomputer was so powerful it tripped the local power substation.

“We knew that we had to re-examine every aspect of the data center infrastructure in order to support our unprecedented power and cooling density. We brought in a fully-custom designed CDU to support dojo’s dense cooling requirements,” Rajiv Kurian, Principal Engineer at Tesla says during the presentation. 

He continues, “Since our Dojo cabinet integrates enough power and cooling to match an entire row of standard IT racks we need to carefully design our cabinet and infrastructure together and we’ve already gone through several iterations of this cabinet to optimize this.”

“Earlier in this year we started load testing our power and cooling infrastructure and we were able to push it over two megawatts before we tripped our substation and got a call from the city,” he adds.

Leave a Reply

Your email address will not be published.

Related Posts
Total
44
Share

Get daily funding news briefings in the tech world delivered right to your inbox.

Enter Your Email
join our newsletter. thank you