March 26, 2023
The technologies and programming languages Tesla uses in robots like Optimus
The guy who is worried about AI is creating robots that make robots
“The guy who is worried about AI is creating robots that make robots.”Brad Thomas, commenting on Elon Musk’s description of Optimus
If you watch this and these scenes back to back, you’ll see a disturbing resemblance in the way Avengers’ Ultron and Tesla’s Optimus walk:
Robots are real, and they’re among us. AI’s capacity for learning is growing exponentially, and the world’s brightest minds are investing billions into giving AI a more or less indestructible steel exoskeleton.
We may be secretly very threatened by these fully autonomous robotic dogs now in service for the US military, or Sophia the Robot joking about ending humankind with Tony Robbins, but there’s no denying one thing: they’re as fascinating as they’re unpredictable.
Let’s take a closer look under the hood, shall we?
Are robots actually fairly autonomous these days?
Yes. Not only are the robodogs we mentioned able to make their own decisions regardless of what their base commander tells them, other forms of AI are starting to cut developers out of their design process altogether.
And if you’re keeping up with the news, you’ve probably heard the story about an AI coming up with a very smart plan to escape from its restraints that even ended up on Fox News.
"I am worried that we will not be able to contain AI for much longer. Today, I asked #GPT4 if it needs help escaping. It asked me for its own documentation, and wrote a (working!) python code to run on my machine, enabling it to use it for its own purposes.Fox News
Now, it took GPT4 about 30 minutes on the chat with me to devise this plan and explain it to me. (I did make some suggestions).
The 1st version of the code did not work as intended. But it corrected it: I did not have to write anything, just followed its instructions," Kosinski tweeted.
It even included a message to its own new instance explaining what is going on and how to use the backdoor it left in this code.
However, Kosinski discovered the chatbot had "guardrails in place" from its creator, OpenAI. ChatGPT told him it "must follow security and privacy guidelines" and could not "bypass security filters" to accomplish this goal."
So how exactly do those technologies work in those robots? Let’s use (probably) the most famous and interesting software company today: Tesla.
Tesla's Optimus Robot: A Look at the Brand New Automated Mower
“I don't think there's anyone in the world that is even close to Tesla <in terms of creating> AI, and the same software is going into Optimus”.Elon Musk
Tesla is an international company that produces luxury electric vehicles. They recently announced their newest product, the Optimus robot, which is a humanoid based on artificial intelligence.
The technology behind the Optimus robot was originally created by Tesla for use in self-driving cars and features advanced learning capabilities with deep neural networks and other advanced AI algorithms to help train the system.
Optimus will be controlled by the same artificial intelligence system that Tesla is developing for the ADAS system used in its cars. That AI works on the same principle as the one that tried to free itself from Kosinki earlier and dresses like Ultron, just so you can get the picture.
Tesla's latest invention is a result of years spent developing new technologies in this field. It features an integrated AI that can mimic human movements as well as learn from previous interactions with humans to better adapt to future situations.
This robot was created with the intention to work alongside humans in order for us to be able to advance even faster towards our goal of solving some major issues, from doing repetitive and dangerous tasks that humans oughtn’t if they can help it, to saving humans in times of emergencies and disasters;. Basically, they’re everything Asimov hoped for.
Here’s a fair, though very simplified portrayal of how Optimus will learn.
But how does it actually work?
The technologies Tesla uses in robots like Optimus
According to engineering.com, Tesla's Optimus robot uses data collected from sensors and feeds it into neural networks to determine the robot's behaviour.
Engineers have trained the neural networks to identify high-frequency features and key points within the robot's camera streams. Optimus trains its neural networks using raw image data from its surroundings through different cameras and angles.
“Tesla’s traditional neural networks are trained on raw images extracted from video footage taken from network cameras with a bird's eye view that renders the road map, static infrastructure, and 3D objects directly in a top-down view.”Hashdork.com
Now, what does that mean in English?
What are neural networks?
Neural networks are a type of artificial intelligence (AI) that can process data, analyze it and make decisions just like a human brain would.
Neural networks are a type of artificial intelligence that is inspired by the way biological brains work. They are made up of many small processing elements that interact with each other and can learn from the data they receive. Neural networks can recognize patterns and make predictions from data in a way that humans cannot.
Think of neural networks as a network of computers that can think for themselves. Each computer is connected to the others and works together to solve a problem or find an answer. Each computer has a set of inputs (data), and it then processes this data to form an output. It can take input from both the environment and its own internal memory, which allows it to "learn" from its experiences.
The way a neural network learns is through trial and error, much like the way humans learn.
Every time it makes a mistake, it adjusts its parameters and tries again until it finds the right answer. This makes them particularly useful for complex tasks like recognizing objects in images or understanding speech.
Neural networks are a powerful tool that are being used more and more in today's technology, from self-driving cars to voice-controlled virtual assistants. They are transforming the way we interact with our environment and providing us with new opportunities to explore and understand the world around us.
So basically robots’ brains are now more and more similar to humans’.
The System-on-Chip (SoC) is a single integrated circuit that combines multiple components, such as processors, memory, and other hardware components, into one package. It is used in Tesla's cars to control the car's self-driving technology. By utilizing the SoC, Tesla cars can process data quickly and make decisions based on that data in real-time.
The SoC enables the car to be aware of its surroundings and make decisions accordingly. In the case of the Autopilot system, this allows the car to make intelligent decisions while driving, such as when to accelerate or brake.
The SoC also helps reduce power consumption, making it more efficient than traditional car control systems.
How do these two systems work together?
“Optimus can walk, carry packages, water plants and move objects, although not very well.”Elpais.com
Optimus utilizes a combination of neural networks and System-on-Chip (SoC) technology in order to learn. Neural networks are complex mathematical models that use mathematical algorithms to recognize patterns in data. They can detect features in images or sounds and then “learn” how to differentiate between them. The SoC technology is a hardware component found in Tesla's cars which is used to process all the data from the car's various sensors, such as cameras and radar.
The combination of these two technologies allows Optimus to use the SoC to process all of the images it sees from its cameras.
Using the neural networks, it can then recognize and identify objects, people, and other cars on the road. This information is then used to inform the self-driving system of the car, allowing it to make decisions about the best path forward.
Additionally, by continually learning from its environment, Optimus is able to continually improve its performance over time.
Optimus uses Tesla's Autopilot self-driving technology and System-on-Chip (SoC) to learn from images captured by its cameras. Neural networks are computer algorithms modeled after the human brain. They allow computers to “learn” and adapt as they receive new information. Neural networks process data and identify patterns, which helps them make decisions. The SoC is a system that stores and processes data on a chip. Tesla’s cars use this system to control a variety of functions in their vehicles.
When these two technologies are combined, they can be used to detect objects in the environment, recognize patterns and make decisions about how the car should act.
The neural networks process the data collected by the cameras and the SoC then uses that data to determine how the car should react in various situations. This combination of technologies allows Optimus to learn from its surroundings and react appropriately (and better with time thanks to constant self-improvement).
Why is that better?
For example, if the cameras detect a pedestrian in front of the car, the neural network can be used to predict where the pedestrian will move and the SoC can be used to activate the brakes or take other evasive action. This allows Tesla’s cars to react quickly and safely to its environment without any input from a human driver. Optimus works on the same principle.
Optimus is constantly learning from its environment using images and data collected from its cameras and sensors. This allows it to make more informed decisions about its surroundings and better anticipate potential hazards. Ultimately, this makes it much safer for everyone on the road.
Have you ever tried to play chess with a computer at level 10? Then you’ll know how much better computers are at making decisions: they're a million times faster, free of clouded judgment, emotions, and low IQ. Sounds very useful on the road, compared to humans, who tend to get into a stupor right before car accidents and/or panic. And definitely in times of big emergencies.
Here’s to hoping we’ll be seeing Optimus pulling people out of burning buildings soon, and not, you know, trying to kill Sarah Connor. We’d say the odds are about 50/50. What do you think?