- At the highly anticipated Tesla AI Day event, the electric car maker made several announcements in the field of computing, AI and robotics.
- The highlights include the Tesla Bot and Dojo D1 chipset.
- The Tesla Bot is a humanoid robot, while the D1 chipset is designed to train AI models in the company’s data centres.
- Apart from this, Tesla also shed light on its full self-driving capabilities and plans for the technology’s future.
announced that Tesla is working on rolling out a prototype humanoid robot by next year. The announcement was made at the
Day event in California, where the company looks to recruit new machine learning (ML) talent.
At the event, Tesla also unveiled a new custom chipset that trains artificial intelligence models at its data centres. The new Dojo D1 chipset, a part of Tesla’s Dojo supercomputer system, has 362 teraflops of processing power. It’s built on a 7nm fabrication process.
Here are the top highlights the Tesla AI Day announcements:
Tesla Bot – the company’s first humanoid robot
Elon Musk, the company’s co-founder and chief executive officer, claimed that Tesla is “basically the world’s largest robotics company”.
The Tesla BotTesla
Musk said that the Tesla Bot will leverage some of the hardware and software used in the company’s autopilot driver assistance programme.
He further stated that the Tesla Bot will be designed to handle unsafe, repetitive and boring tasks, essentially helping the company solve the problem of labour to some extent.
“I think essentially in the future, physical work will be a choice, if you want to do it you can,” Musk said.
Tesla’s website also reveals that the company is scouting for talent to help bring the Tesla Bot to reality.
“We’re seeking mechanical, electrical, controls and software engineers to help us leverage our AI expertise beyond our vehicle fleet,” the company stated on its website.
Dojo D1 chipset to train AI models
Tesla unveiled a new chipset named D1 to train the AI models used in the company’s data centres. This, Musk says, will allow the Dojo supercomputer to process imaging data four times faster than other computing systems.
The new Dojo D1 chipTesla
The underlying idea here is to push out the AI models trained by the D1 chipset to existing Tesla cars via software updates.
Another aim of building the D1 chipset and having the hardware and software vertically integrated in-house is to reduce the latency and improve bandwidth, resulting in better AI performance.
“We can do compute and data transfers simultaneously, and our custom ISA, which is the instruction set architecture, is fully optimized for machine learning workloads. This is a pure machine learning machine,” said Tesla director Ganesh Venkataramanan.
Simply put, the new D1 chip is faster, has more bandwidth and will help Tesla take a leap in AI performance.
Musk said the Dojo supercomputer system should be operational next year.
“Full self-driving” to get better this year
Tesla currently sells a “Full Self-Driving Capability” add-on that costs $10,000. This add-on lets the Tesla cars automatically change lanes while driving, navigate on highways, park the car or emerge from a parking spot.
Now, Tesla says that later this year, this package will get an additional ability to automatically steer on city streets.
Tesla is also offering the add-on for a monthly subscription of $199 for those who don’t want to pay the lump sum of $10,000.
Expanding self-driving to other cars
While elaborating on the current status of full self-driving, the company hoped that the learnings from the Dojo supercomputer system in the field of self-driving will not be limited to just Tesla and that it could be expanded to other carmakers down the road.
“This is not intended to be just limited to Tesla cars. Those of you who’ve seen the full self-driving beta can appreciate the rate at which the Tesla neural net is learning to drive,” Musk said.
The full three-hour event can be watched below: