Elon Musk unveiled The tesla bot ‘Optimus last month at the company’s AI Day. Last year He teased the Tesla robot at the same event by calling an actor in a suit to come on stage, but this time the real robot walked onto stage by itself, unconnected to any power source or stabilizer for the first time.
The goal, Musk said, is “a fundamental transformation for civilization as we know it” at what Musk guessed would be a very reasonable cost: under $20,000. (Take that with a grain of salt: Tesla cars were supposed to be $35,000 once upon a time as well.) Widespread production, however, could lead to the “end of poverty,” according to Musk, and lead to an economy that becomes “quasi-infinite.”
Musk also stated that Optimus will someday be more significant to Tesla’s business than the cars that currently provide the majority of the company’s revenue.
Here’s what we know about Optimus, so far.
- Brain: Tesla AI chip
- Hands: 11 degrees of freedom
- Muscles: 28 structural actuators
- Joints: inspired by biological human joints
- Eyes: cameras
- Ears: microphone
- Voice: speaker
- Battery: 2.3kWh battery pack, 52V
- Power consumption: 100 watts while sitting, 500 watts while walking
- Speed: 5 MPH (8 kilometers/hour)
- Connectivity: WiFi, LTE
- Weight: 161 pounds (73 kilograms)
- Carrying capacity: 20 pounds per hand (9 kilograms), likely more in different configurations
- Materials: metal where necessary but as much plastic as possible for weight savings
In terms of movement capability, the Tesla robotics team is aiming extremely high, with no fewer than 18 sometimes-complex movements. Note, these are Tesla-reported, and the robot almost certainly does not do all of these yet.
- Forward walking
- Squatting and squat walk
- Turning while walking
- Lifting objects from the ground up to eye level
- Squeezing or gripping an object and lifting it
- Climbing stairs
- Squatting and picking up an object
- Walking on a slope or hill
- Sliding objects
- Using a drill
- Pushing and pulling objects
- Turning with an object
- Using a screwdriver
Now let us look into the hardware of the Bot
The Impressive Battery
The battery is impressive for being able to power the robot for a full day (ish). Again, until Tesla does this, it’s not all that relevant, but Tesla knows a lot about power systems and batteries, and we can be sure that the firm will be able to deliver on its promise.
Safety Of Robot
We should admire that Tesla is thinking about how to frame its robot such that it may fall safely and get back up with only minor damage though the company doesn’t appear to be using any form of a protective motion for fall mitigation, which is an important field of research elsewhere. And the safety of others is not discussed in this situation. We can be happy the robot won’t be too badly hurt if it falls, but can Tesla say the same for anyone standing next to it?
Tesla’s bespoke actuators appear to be fairly reasonable. Not noteworthy, but Tesla must manufacture its actuators if it requires a large number of them, which it allegedly will. Given Tesla’s degree of mechanical ability, we can expect these to be completely adequate, but nothing here is crazy small, cheap, efficient, powerful, or anything else. And it’s difficult to judge how well the actuators will perform just on these slides and the presentation, especially for dynamic motions. The robot’s software still has a lot of catching up to do.
Tesla used six cable-driven actuators for fingers and thumbs (with springs to supply the opening force) for simplicity and to reduce part count. This is maybe surprising given that cable drives are often less durable and more difficult to calibrate. Tesla claims that the five-finger hand is required since Optimus will be operating with human tools in human situations. It is certainly one viewpoint, however, there is a significant trade-off in complexity because the hand is made to hold a 9-kilogram bag.
Now let us move on to the software part of the Optimus.
A Walking Car?
The following quote comes from Milan Kovac, who was on the autonomy team during the presentation.
“All those cool things we showed earlier in the videos were possible in just a few months, thanks to the amazing work we’ve done on Autopilot over the past few years. Most of those components ported quite easily over to the bot’s environment. If you think about it, we’re moving from a robot on wheels to a robot on legs. Some of the components are similar, and some others required more heavy lifting.”
I am not sure about the implication that “humanoid robots are just cars with legs,” but it’s impressive that they were able to port much at all I was skeptical of that last year, but I’m more optimistic now, and able to generalize between platforms (on some level) could be huge for both Tesla and autonomous systems more broadly. Though I’d like more information on what was simple and what was difficult.
When Will It Hit The Market?
The timeline for the mass production and launching, like all Elon Musk timetables, is extremely ambitious. Tesla hopes to start sales within three years, and most likely before five years. Given the task’s intricacy, this is difficult.
“Our goal is to build a viable humanoid robot as soon as possible,” Musk explained.
He was unafraid to foresee the repercussions of fully functional robotic workers and helpers, claiming that they will lead to a “future of abundance, a future without poverty, a future where you can have whatever you want in terms of products and services.” Musk admitted that there are robots out today that look to be more capable, maybe from companies like Boston Dynamics but stated that Optimus, unlike them, is intended for mass production: millions of units.
Achieving that would result in a fundamental shift of civilization as we know it, according to Musk.
He is not mistaken if Tesla or any other company can develop workable, functional, and useful robots that are widely available at a low cost, it will revolutionize nearly impossibly large areas of current economies and societies. High-level humaniform robots could do tasks such as warehouse workers, fast food workers, janitors, factory workers, construction workers, cleaners, gardening workers, shipping and receiving staff, and stocking clerks.
However, roboticists doubt Tesla’s ability to deliver them anytime soon.
Diana Rus, an MIT professor and director of the Computer Science and Artificial Intelligence Laboratory, recently shared that “the more you generalize, the less you optimize,” which is why so many robots today do not take human form. One way of addressing that is reconfigurable robots.
“There is a kind of a trade-off between how effective the robot is at doing a set of tasks and how many tasks the robot can do,” Rus said. “My idea was to create universal robot cells that could combine to form different types of machines … so the shape and the function of the machine would be specialized, but since each machine will be built out of the same building blocks, we could have a kind of a generality towards how we think about robots.”
This is known as modular self-reconfiguring robot systems. Time will tell if those types of robots, or specific-purpose robots, or Elon Musk’s general-purpose humanoid robots, will triumph.