Tesla takes Optimus to the next level!
If you weren't paying attention to what is happening in the world of humanoids, you are excused and justified because, for decades, this has been a gimmick of research labs and not something attainable for the common human.
Two years ago, that changed.
Before TESLA moved into this space, all those robotics efforts were underfunded, underscored, and essentially research playgrounds.
If you missed the previous episodes on the subject, here is the link for you.
Tesla introduced an idea. A humanoid that can help get stuff done. First in the factory and eventually at home. In the long run, in Space. So, in terms of ethics, workers are not the only ones sweating for their jobs shifted to AI and robots. Even the NASA rover on Mars needs help sleeping at night since the news broke.
I want to underscore that it was a concept to the degree that they used a human in a costume to inspire what they were going to do. Granted that the dude did a nice dancy act, we are still talking that they were in hiring mode, not in "the pie is in the oven" mode.
Musk, as a well-known talent in scouting and uncovering talent pools, used the event to further spot additional talents that could have made the vision a reality. And the robotics team at Tesla is definitely displaying that talent eye. A few months after that dance, they unveiled:
Musk has a talent for picking businesses that are difficult but attainable, for finding engineers up for the challenge, and providing them an environment where they expect to make a difference. --CNN
Bumblebee is the nickname for their first prototype. It couldn't stand on its own when it was presented on stage; it needed a metal platform to hold it in place. Remotely, they could control the limbs, but that was IT in terms of "we got a bot." It was also built with off-the-shelves parts. It did look good though!
Well, talent and passion are the perfect combo for accomplishments that are unique in their own merit. Just a few months after Bumblebee was presented, Optimus Gen 1 was born.
They were using their own actuators, the design overall had become more sleek and uniform to the presented concept in 2021.
In September 2023, they broke the internet in presenting Optimus Gen 2, that demonstration almost touched the yock, of what they were after. That is:
Build a humanoid that is useful for some tasks, without a pre-programmed workflow but rather with a general intelligence that allows to robot to observe a human doing a specific task and then repeat that task accordingly. That is not "playback" what a human did, but rather emulate what the human performed and adapt if something not seen during the training occurs.
As an example:
A human shows the robot how to pick a screw from a box, grab a screwdriver and turn the screw into a hole.
The robot repeats what has observed. However, in the process of picking up the screw, the magnetic effect of the screwdriver fails to hold the screw and the screw falls on the workbench. In a pre-programmed scenario a robot would screw an invisible screw because it doesn't know what happened. In the general intelligence scenario Optimus detects that something unexpected happened and it handles the scenario gracefully. E.g. pick up the fallen screw or pick a new one from the box.
In terms of robotics this an absolutely challenging operation to compute and address, just five years ago this would not even have been a discussion to have for a humanoid produced at scale.
To put it in contrast, Gen 1 was screwing mechanical parts, Gen 2 can fold t-shirts after it has been shown by a human on how to do it!
Electro-mechanical innovation is at afoot for many companies, building a digital brain for those humanoids to operate is a much narrowed scoped opportunity. And Tesla with their massive investment in AI over the years is a tough bar to meet.
Boston Dynamics have been trying around this space for almost two decades and yet because the lack of a digital brain investment, they have a dog as their major accomplishment in 15+ years of operations and not a humanoid. That's why we particularly appreciated musketon artwork on the matter.
Sure enough, Boston Dynamics Atlas, their development platform for humanoids has been showing flipping and dancing skills but those are pre-programmed choreography acts, there's no watch and learn training and they have no manufacturing facilities like Digit to bring innovation forward on the path of humanoids.
Digits uses pinchers while most multi environment humanoids use hands. A frequent question we receive via our social outlets and paid programs is: why are they trying to build those robots in a human form?
the most straightforward answer I can offer this:
humans built their environment according to how they move and interact with objects and spaces. A door knob or handle is designed for a human hand. The distance between desks in a office or the space in a bathroom stall is built for a "standard human size".
That means, the height, body size and limbs have to resemble a human if we want those bots to circulate in human spaces and operate things designed for humans.
Below you find a comparison table between the two generations of the Tesla Optimus Prime.
| Category | Gen 1 | Gen 2 |
|---|---|---|
| Appearance | White and black color scheme, screen on head displaying a face | Sleeker, futuristic design, silver and blue color scheme, transparent visor on head showing eyes, more visible joints and wires |
| Size and Weight | 5 ft 8 in (173 cm), 125 lb (57 kg) | 5 ft 10 in (178 cm), 130 lb (59 kg) |
| Capabilities | Walk, wave, dance on stage, limited mobility and dexterity | Freer arm, hand, and finger movement, can sort colored blocks, locate limbs in space, maintain yoga pose, same carrying capacity |
| Status | Proof-of-concept prototype, shown at Tesla’s AI Day event September 2022 | Semi-functional prototype, displayed at Tesla’s AI Day event September 2023, production-ready version expected by 2024 |
If you want to see them in a side by side comparison, just play the video below to get a good sense of the massive progress in such a short period of time across two generations of the bot. Remarkable progress by any means!
It'll be a fundamental transformation for civilization as we know it
said Musk. He continues to say that Optimus has the potential of "two orders of magnitude" of potential improvement of economic output.
The Ethical Debate
There's no question that at some point humanoids, in particular, will be displacing human labor. There's also no doubt on the principle that who can't assist their parents, for distance or time, will happily rent a bot to fulfill caregiver duties. Who feels lonely, will find comfort (and pleasure...) in adopting a companionship. All those changes at this point are postponed only by the affordability factor.
Our brains and cultural dynamics will be put to a strain. Corporation will want to adopt and invest as much as they can to reduce the human element in their operations. No healthcare plans, no vacations to plan around working schedules, from a purely capitalistic point of view, for corporates where repetitive tasks and reliability are a factor in success, humans are the bottleneck and a higher cost.
As people age, they may exhibit behaviors similar to children, such as more direct emotional expression, increased dependence on others, and a simpler enjoyment of life, possibly due to cognitive, physical, and social changes.
In this context, Robot & Frank movie are more real than ever!
If today someone is taking care of your aging parents, tomorrow, you may consider to adopt/rent a bot to be there full time and a much affordable price. And all of the sudden now we face what I call the "toilet paper moment".
The bots are coming and they will do an incredible amount of good, at scale. Very likely over decades most families will have at least one but their presence will stress out the existing laws, social norms and appear in the least expected environments.