Today, a robotics startup called Figure is unveiling “the world’s first commercially viable general purpose humanoid robot” called Figure 01. Shown in the rendering above, Figure 01 does not yet exist, but according to this morning’s press release, it “will be able to think, learn, and interact with the environment and is intended to be used initially in the workforce to address labor shortages and, over time, help eliminate the need for unsafe and unwanted jobs.” Sounds great when (or if) it happens.

We tend to be skeptical of announcements like this when a company comes out of the shadows with ambitious promises and some impressive renders, but little real hardware to showcase alongside them. What caught our attention in the case of Figure is their exceptionally skilled robotics team, led by CTO Jerry Pratt. Pratt spent 20 years at the Florida Institute for Human and Machine Learning (IHMC), where he led the runner-up team in the DARPA Robotics Challenge finals. Working with DRC Atlas, NASA’s Valkyrie and, most recently, Nadia, IHMC has established itself as a leader in robot design and control. And if anyone is going to turn a useful humanoid robot from an engineering concept into a commercial reality, it’s the people who will make it happen.

Figure was founded in 2022 by Brett Adcock, who also founded Archer Aviation, which has successfully built and is currently flight testing the eVTOL commercial passenger aircraft. The company started out self-funding but will close Series A in the next couple of weeks. Over the past year, Figure has hired more than 40 engineers from institutions such as IHMC, Boston Dynamics, Tesla, Waymo and Google X, most of whom have significant robot experience. -humanoids or other autonomous systems.

“We think this is the best team of humanoid robots,” says Adcock. IEEE Spectrum. “Cumulatively, the team probably built 12 basic humanoid robots,” adds CTO Jerry Pratt. “We’ll have expertise in just about every part of the thousands of things you need to do for humanoids.” Pratt says Figure didn’t originally intend to use a lot of new technology in his robot – it’s not based on some secret drive technology or anything like that. “It’s going to be a new design with really solid engineering.”

The robot that Figure is working on is a “commercially viable general purpose humanoid” that would look something like this:

Obviously the above video (and all the robot images in this article) are renders and do not show the actual robot performing real actions. However, these renders are based on a CAD model of an actual robot that Figure plans to build, so Figure expects their final hardware to be very similar to what they show today. Which, if it ends up being that way, will be impressive: it’s a very thin form factor that puts some limits on its performance. The latest robot will be fully electric, 1.6m high, weigh 60kg with a payload of 20kg and will run for 5 hours on a single charge.

Three engineers are working on a model of a black humanoid robot.Figure

“Having a humanoid form is very difficult to make packaging,” explains Pratt. “In general, with the technology available today, you can achieve somewhere between 50-60% of most human characteristics, such as degrees of freedom, peak speeds and torque and the like. It won’t be superhuman; we will focus on real applications and not try to push the boundaries of pure performance.” This focus has helped Figure limit its design in pursuit of commercial utility: you necessity the robot must be thin in order to work in rooms intended for people. With this design philosophy, you won’t get a robot that can do back flips, but you will get a robot that can be productive in a cramped workspace or walk safely through a crowded warehouse.

This is due to the reason Figure builds a humanoid robot in the first place. The added complexity of the legs must somehow be justified, and Figure’s point of view is that building a robot without legs that have the necessary range of motion to do what they need to do in a human workspace would be complex enough that you could just build a robot with legs anyway. And this opens up an opportunity (or perhaps an imperative) for generalization. “If you’re building humanoids, you pretty much need to find a general purpose,” says Pratt. “For just one application, there will probably always be a specialized robot that is better.”

“Whenever people say, ‘Why are you making humanoids, why don’t you do something better?’ Maybe we’re just limited by our imagination, but I can’t imagine what it could be. With today’s technology, it is impossible to get close to a person, so I think that the strategy of being as close to a person as possible is quite justified. — Jerry Pratt

The figure, like most other companies working on commercial humanoids, sees warehouses as an obvious entry point. “The warehouse makes it easy for us,” says Adcock. “It’s indoors. There are no clients around. AMR already exists [autonomous mobile robots] and cobots [collaborative robots] work around people. And there is a warehouse management software system to manage high-level behavior. We’re betting that if we can figure out how to get one big enough application and deploy enough robots, we can add new software as we do more stuff and over time produce really big volumes and make robot affordable. “. Adcock acknowledges that the robot must make financial sense in the market it targets. That is, if it is going to replace human labor in the warehouse, it must be cost-competitive with human labor, which will be a major problem that may (at least initially) rely on some form of human telecontrol for maximum reliability. .

Figure believes it has a realistic chance of being the first company to actually commercialize a general purpose humanoid robot, although both Adcock and Pratt have indicated that there is so much potential demand that they are not particularly worried about competition. “I think it’s just a matter of how to get there,” Pratt tells us. “There is room for several companies and I think we can be one of them.”

I don’t think anyone would dispute the existence of general purpose humanoids. I think the only question is when they will happen and what it will look like. — Brett Adcock

To achieve this, as Figure makes clear in its master plan, “major technological advances will be required.” Here is what the company believes is needed for this:

  • System hardware: Our team is developing a fully electromechanical humanoid, including arms. The goal is to design equipment with the physical capabilities of a non-expert. We measure this in terms of range of motion, payload, torque, cost of transport and speed, and will continue to improve with rapid development cycles, each cycle as part of a continuum.
  • Unit cost of production: We aim to reduce unit cost per humanoid through high-volume batch production while working towards sustainable economies of scale. We measure our costs in terms of fully charged operating costs per hour. I believe that with a high rate of mass production, the cost per unit of production will fall to an affordable level.
  • Safety: It is essential that our humanoids can safely interact with humans in the workplace. We will design them to meet industry standards and corporate requirements.
  • Volume of production: We foresee the need not only to supply a high quality product, but also in exceptionally high volumes. We expect a steep learning curve as we move out of prototyping and into mass production. We prepare for this by taking care of design for production, system safety, reliability, quality and other production plans.
  • Artificial intelligence: Creating an artificial intelligence system that allows our humanoids to autonomously perform everyday tasks is perhaps one of the most difficult challenges we face in the long term. We solve this problem by building intelligent agents that can interact with complex and unstructured real world environments.

This all sounds very convincing, but it is important to note that, to the best of our knowledge, Fig. made any of this yet. They have goals and objectives, they design according to those goals and objectives, and they can do their best to anticipate and anticipate some of the challenges ahead and plan and prepare as much as possible. However, at this point it’s premature for us (or anyone else) to judge whether the company will be successful as they still have a lot to figure out. To be clear, I believe Figure believes he can ultimately do what he says he plans to do. My criticism here is mainly that the company says more than it shows – historically this hasn’t been a good strategy for robotics, which tends to be vulnerable to broken promises.

The figure admits that this will be a very, very difficult process and that the company faces “high risk and extremely low chances of success”, which is an attractive statement among what would otherwise be quite a lot. uniformly positive hype, for lack of a better word. And Adcock understands that higher targets (like “consumer household” and “off-world” applications) are likely to take some time, telling us the company is “really excited about the potential here for a few decades.” long period. ”

“Part of the bet we’re making is that we’re not too early and that the technology exists for that.” — Jerry Pratt

Torso-up rendering of a black humanoid robot with a black faceplate that can also display information.Figure

So, what is the actual state of the Figure robot now? “We just finished our alpha version,” says Adcock. “This is our first full scale robot. We are building five of them. We hope that he will begin to take the first steps within the next 30 days. And now we have started the second generation hardware and software version, which we will complete this summer.” These are tight deadlines and Figure hopes to develop a new major version of both hardware and software every 6 months, indefinitely. “We think we’re in a good position,” Adcock continues. “We hope that this year we will achieve great success and be the first to enter the market. We’re going to try. We will move as fast as we can to achieve this goal.”