Tesla AI Day Reveals What Is Coming
The Tesla AI Day has been another demonstration of the company’s commitment to further innovate. The carmaker, known for their expertise in EV (Electric Vehicles) and renewable energy systems, have announced what they have been working on in relation to AI (Artificial Intelligence). The presentation began discussions about the implementation of neural networks to the Autopilot system for FSD (Full Self-Driving) on their cars, the DOJO compute hardware used for training and simulating self-driving, advanced 4D labeling used in computer vision for autonomous driving and edge cases for complex scenarios self-driving cars could face.
One of those revelations have sent shockwaves around the tech industry, firing the imagination and igniting interest. This was the Tesla Bot, a robotics prototype machine that will have human-like characteristics. That is probably the best way to describe a “humanoid”, if that was its intended purpose. You have the top EV carmaker planning to build robots using their AI driving technology, in a human-like form. This is like how Boston Dynamics built their robot (i.e. Spot) that mimics a dog. A robot that mimics a real human would be raising the bar in advanced tech.
According to a report from The Verge:
“The humanoid robot will leverage Tesla’s experience with automated machines in its factories, as well as some of the hardware and software that powers the company’s Autopilot driver assistance software.”
The Tesla Bot
According to Tesla, the Tesla Bot prototype model will be a bipedal 5' 8" (height) with a weight of 125 lb. It will have a carrying capacity of 45 lbs. and deadlift up to 150 lbs. In terms of motion, the robot can walk or travel about 5 MPH (Miles per Hour). The robot will be built for general utility purposes.
It will be built from light-weight materials (exact materials not verified). This is for flexibility and maneuverability, assuming that will allow the robot to move smoothly with less effort and power requirements. From the presentation, Tesla shows a slender humanoid form that appears like a mannequin. It will have mechanical actuators and sensors all around including a visioning system consisting of 8 cameras powered by Tesla’s own hardware.
This is part of Tesla’s push for AI in general purpose robotics combined with AI and computer vision. This would allow the robots to act autonomously to perform certain tasks. The other technical details regarding the power system, locomotion, dexterity and mechanics were not discussed. This gives a rather more general idea about the prototype before it launches.
The Purpose
The prototype robot Tesla (code named Optimus) is planning to build will be like an EV with two legs. It is going to use the FSD visioning and AI capabilities of Tesla’s EV with a humanoid form. The purpose of the robot will be:
- Eliminate dangerous repetitive, boring tasks
- Serve as a utility or backup to personnel
- Provide a non-threatening friendly robot to assist humanity
There are many tasks out in the world that can affect human well being. This includes hazardous worksites, where dangerous chemicals and elements are found (e.g. chemical plants). These tasks can also be boring and can cause more stress that affects human health. Many jobs in industries like automobiles have already used robots for automation. More jobs can be replaced with a human-like robot. It may also fill in jobs that many humans don’t want to work in. This includes sanitary work, janitorial services and industrial cleaning.
Such robots can also help with dangerous field work like law enforcement. A robot can help collect information on the ground, to help with sting operations. They can also enter suspected areas to gather data, where it would be less safe for a human operative. These robots can also provide backup to help laborers working in construction or assist medical staff in hospitals. We can point out that robots can help health workers since they won’t be at risk for viral infections or diseases.
These robots can also provide comfort and assistance to those with special needs. The presentation never discussed this topic, but it appears to have a valid purpose. Robots can help rehabilitate disabled or injured patients. They can also help like caregivers. At first the robots would be limited to their tasks, but as they evolve their capabilities could increase. They can fill in the need for caregivers, but providing the emotional and psychological response is going to be the challenge.
The Takeaways
While ideas like these seem incredible and straight out of science-fiction, they later turn up to have use-cases in real world applications. Some ideas during the dot.com period that failed (e.g. grocery delivery, online music etc.), are now normal. They just take time to evolve. The same can be said about humanoid robots, but there are also concerns. Imagine a humanoid with a mind of its own allowed to walk the streets.
There is the fear that AI could take over the world. Even Tesla’s CEO Elon Musk was fearful about AI, since it has the potential to lead to global catastrophe. He is for regulation when it comes to the AI space. The Tesla Bot does fall under this category so it seems odd that he is now developing something that could also have dire consequences. Musk has nothing against AI itself, but the agendas that could be behind its development. The worst case scenario is that the AI goes out of control and that can be like a “Skynet” moment (in reference to the movie Terminator) for these robots.
Should robots be given a mind of their own? That is the issue I think this project will face. If the AI is controlled in some way to make sure the robots only perform certain tasks, it is manageable. When it becomes more autonomous in what it can do and how it can react, then the issue becomes whether they will be a threat to society.
Robots don’t have emotions, but based on the AI there is a possibility they will try to become as human-like as possible. Unfortunately that also means they can develop human reactions that can sometimes lead to the wrong decisions. It is probably best (in my opinion) to have the robots programmed to be specific and not too general. There is a theory that if the robot learns it is better than any human, it may then act on that belief and this is where things can go wrong.
Companies will probably have to be more transparent about the purpose of their AI. While the military will make use of it to their advantage, there is also likelihood in state sponsored AI that can violate human rights. Take for example surveillance systems that implement advanced AI technology (e.g. face recognition). The technology behind AI can always be abused, which is why there needs to be more transparency behind the projects that implement it. The last thing the world needs are robots manipulated by rogue actors to seize any type of authority.
FSD is also not proven highly reliable for self-driving cars, much less for robots. While a prototype can be built, my assumption is that it will not be ready for production use until FSD is proven safe and reliable to regulators. I see the point of just testing Tesla Bots in a lab setting, where its developers can closely monitor and train it.
Synopsis
Tesla is really proving themselves to be more than just a car company. The AI Day event shows that they are up to more things than critics could have imagined or thought of. It was a showcase of computer vision and AI that would satisfy any engineer’s curiosity. The critics might continue to say that Tesla is just on some grandstanding platform showing off things that wastes money and never become reality. If that was the case, we would not have seen the Model 3, Power Wall or Autopilot technology at all.
The Tesla Bot is now the next subject of ridicule for critics. It is just too early to write off anything Tesla announces. What they announced during Battery Day was revolutionary to the EV industry. The AI technology that Tesla has been working on is on a high level, but the average person will not understand it as well as scientists or engineers. I am more under the belief of judging something when it is there, not while it is still being conceptualized.
It took millions of years for nature to build a bipedal species i.e. humans. There is more complexity for Tesla, since bipedalism cannot be easily replicated with technology. This is why not much should be expected at first, from the prototype. They will have to do what nature has perfected in humans along with the dexterity, intelligence and kinematics in robots. A feat like that is definitely going to be an accomplishment until something terrible happens. That is why the issues and concerns also need to be addressed if humanoids will one day walk among humans. It should be about disruption and not destruction.
