Tesla Ai Day: Elon Musk announces an amazing robot and puts on the autopilot

This Thursday, Tesla organized its first AI Day, dedicated to the development of artificial intelligence. Elon Musk's firm spoke at length about Autopilot and presented Tesla Bot, a surprising robot.

Tesla's Autopilot is now popular in the automotive industry. The semi-autonomous system of the American brand is one of the most successful in circulation and the manufacturer is betting heavily on it.

With FSD, Full Self Driving, this will eventually allow cars to drive without human assistance. While Musk recently explained that developing the technology was more complex than expected, the brand detailed the next steps in its development during its first AI Day. A priority for the boss who is betting on it more than ever.

Tesla details the development of Full Self Driving

Andrej Karpathy, the company's director of artificial intelligence, gave an explanation of Full Self Driving. He clarified in particular that Tesla was seeking to reproduce the analysis of the human brain in the face of the situations encountered.

Karpathy in particular revealed that one of the big challenges for Tesla is to merge the eight cameras used for Autopilot. Indeed, these sometimes display the same vehicle and can lead to an error in perception of the environment.

For this, the system is trained to calibrate the visuals of the various cameras and to modify them to make a coherent whole. This operation makes it possible in particular to use a maximum of images around the cars for more relevant decision-making.

Another issue with Autopilot is its memory. It fails to remember road conditions to predict, for example, the direction that other vehicles will follow. The engineers therefore train him to memorize the signs and markings, so that the system can more easily envisage the sequence of events.

When a difficulty arises in traffic, Tesla reproduces it in his simulations, seeks to understand it, and eliminates it. To accelerate the development of the FSD, the manufacturer has already processed 371 million images.

Tesla Bot, the robot derived from Autopilot

After several tens of minutes of conference, and before questions from the media, Elon Musk made an intervention which was not unlike those of Steve Jobs during the Apple keynotes of the time.

Like Jobs’ famous “one more thing…” (“one more thing”), the boss made a surprising announcement. He thus declared that “Tesla is much more than an electric car company” and announced a brand new product.

This is none other than the firm's first humanoid robot, soberly named Tesla Bot. This unexpected project is in development and could evolve from next year to the prototype stage.

Measuring 1m72 and weighing 56 kilos, he can help with common tasks. With a maximum speed of 8 km/h, it will be able to carry loads of up to 20 kilos and lift 68 kilos.

The link with Tesla electric cars is easy to find, since the technologies used will be similar. The Tesla Bot will thus embed the FSD chip present in the Tesla, and will be guided by the same cameras that guide the Autopilot. These will be located in the head of the robot.

This technology will allow him to evolve in an environment with moving people and machines, such as a house or a business. Tesla plans to train the Tesla Bot for its tasks using the Dojo supercomputer, also currently in development, and expected by Musk in 2022.

Depending on the boss, this robot can perform dangerous, repetitive or boring tasks. To move in the most human way possible, it will carry 40 joints.

Internally, the code name for this humanoid is Optimus. Musk wants to make it a force capable of working with humans in the future. But AI Day, both on the Autopilot and on the Tesla Bot, had another interest for Tesla.

A technological platform for recruiting

These various presentations also aim to recruit engineers. We expected it before the event, and the brand confirms it.

At the end of the conference, the tesla.com/ai website was launched. There are various announcements where the manufacturer reveals the areas in which it needs new blood.

Thus, on the material level, he is looking for people who can create electronic chips and develop them. The neural networks department also needs engineers who know how to work on this type of network.

Similarly, Tesla wants employees to work on autonomy algorithms, code underpinnings, and evaluation infrastructure. These three tasks are used to optimize and verify the operation of software systems, while the brand also recruits engineers to work on the Tesla Bot.