You can easily walk the entire Auto floor without seeing a single humanoid. There were a total of three, by count – or, rather, three prototype units that didn’t work. Neura shows off the promised 4NE-1 robot, amid a more traditional form factor. There is a small photo setup where you can take a selfie with the bot, and that’s it.
Notably absent from the annual Association for Advancing Automation (A3) event was the Agility booth. The Oregon company made a big show at last year’s event, with Digit’s small army moving bins from tote walls to conveyor belts several meters away. It wasn’t an elaborate demo, but just seeing bipedal robots working together was still quite a spectacle.
Agility’s head of product, Melonee Wise, said the company chose to sit this one out, because it now has all the orders it can handle. And that’s what this trade fair is all about: manufacturers and logistics companies are shopping for the next technology to stay competitive.
How big a role humanoids will play in that ecosystem is probably the biggest question on everyone’s mind right now. In the midst of the biggest robotics hype cycle I’ve ever seen live, many are still scratching their heads. After all, the notion of a “general purpose” humanoid robot flies in the face of a decade’ worth of orthodoxy. The notion of a robot has been a fixture of science fiction for years, but the reality is that it is a single-purpose system designed to do one job well.
Despite not having a physical presence, the humanoid subject looms large in the event. So, A3 asked me to moderate the panel on the subject. I admit I initially balked at the idea of ​​an hourlong panel. After all, what we do at Disrupt usually runs 20 to 25 minutes. At the end of the conversation, however, it was clear that we could easily fill another hour.
This is due, in part, to the fact that the panel is — as one LinkedIn commenter put it — “stacked.” Along with Wise, I was joined by Boston Dynamics CTO Aaron Saunders, Apptronik CEO Jeff Cardenas and Neura CEO David Reger. I kicked the panel off by asking the audience how many in attendance would consider themselves skeptical about the humanoid form factor. About three-quarters of those present raised their hands, which is more or less what I expect at this stage.
As for the A3, I’d say it’s entered a cautiously optimistic phase. In addition to hosting panels on the subject at Automate, the organization is holding a Humanoid Robot Forum in Memphis this October. The move marks the launch of the A3 Autonomous Mobile Robot (AMR) Forum in 2019, which presages explosive growth in warehouse robotics during the pandemic.
Investors are less measured in optimism.
“One year after we set our initial expectation for global humanoid robots (total addressable market) of $6bn, we raised our TAM 2035 forecast to $38bn as a result of a 4-fold increase in our shipment estimate to 1.4 million units faster. The path to profitability in 40% red in the bill of materials,” Goldman Sachs researcher Jacqueline Du wrote in a report published in February. “We believe the revised shipment estimate will cover 10%-15% of hazardous, hazardous and automated manufacturing roles.”
However, there are many reasons to be skeptical. The hype cycle is hard to navigate when you’re in the middle of it. The amount of money now changing hands (see: The latest raising figure of $675 million) gives one pause in the wake of various startups collapsing in other fields. This is also the case as robotics investment slows down after a few hot years.
One of the biggest risks at this stage is overpromise. Every piece of new technology runs this risk, but humanoid robots are the lightning rod for this stuff. Just as eVTOL proponents see the technology finally delivering on the promise of flying cars, the concept of a personal robot servant seems within reach.
The fact that these robots look like us leads many to believe that they can – or will soon be able to – do the same things as us. Elon Musk’s promise of a robot that works in the Tesla factory all day and then comes home to make you dinner added fuel to the fire. Tempering expectations is not really Musk’s chapter, you know? Meanwhile, others have been circling the notion of general intelligence for humanoid robots – something that may not be possible (“five to 10 years” is a time frame I often hear).
“I think we have to be careful about the hype cycle, because eventually we have to deliver on our promise and potential,” Cardenas said. “We’ve experienced this before, with the DARPA Robotics challenge, where there was a lot of excitement, and we hit the reality that came out of that.”
One source of disconnection is the question of whether the system is deliverable now. The answer is murky, in part because of the nature of the partnership announcement. Agility announced working with Amazon, Apptronik with Mercedes, Image with BMW and Sanctuary AI with Magna. But every partnership so far should be taken for what it is: a pilot. The exact number of robots deployed in a given partnership is never disclosed, and the number is often in the single digits. It makes perfect sense: These are all operating factories/warehouses. It will be wildly disruptive just slot in the new technology in size and hope for the best.
Pilots are important for this reason, but they should not be mistaken for market fit. As of this writing, Agility is the only group that has confirmed with TechCrunch that it is ready for the next step. In a panel discussion, Wise confirmed that Agility will announce specifics in June. Cardenas, meanwhile, stated that the company is planning a pilot in the “back half” of 2024, with plans to extend it to the beginning of next year.
Neura and Boston Dynamics are only very early stages of the conversation. Neura promises to show some demos at some point in July, moving the 4NE-1 beyond what has now become a series of pre-made videos, plus a unit that doesn’t work in Automate.
While we will see more of the electric Atlas beyond the 30-second video, Saunders said, “(the video) is only meant to be an early peek. We plan to enter the pilot and some more pragmatic pieces next year. Until now, we are mainly focused on building the focus and technology .There are a lot of difficult problems to solve in the AI ​​space. Our team is working on it now, and I think that as these features get stronger, we will have more to show off.
Boston Dynamics didn’t start from scratch, of course. After more than a decade of Atlas, the company has humanoid expertise like no other, while the launch of Spot and Stretch has taught the company how to commercialize a product after decades of research.
So why has it taken so long to see a company swing into the commercial humanoid category? “We wanted to make sure we understood where the value was placed,” Saunders said. “It’s very easy to make a demo video and show good stuff, but it takes a long time to find the ROI (return on investment) case that justifies the human form.”
Neura has the most diverse portfolio of companies on stage. In fact, one gets the sense that every company is finally ready to start humanoid in earnest, it will be just another form factor in the company’s portfolio, rather than a driving force. Meanwhile, when the electric Atlas finally launches, it will be Boston Dynamics’ third commercial product.
Since Digit is an Agility-only offering, the company is very committed to the bipedal humanoid form factor. For its part, Apptronik splits the difference. The Austin-based company has taken the best approach to form factor. If, for example, legs are not needed for a certain environment, the company can attach the upper half of the robot to a wheeled base.
“I think at the end of the day, it’s about solving problems,” Cardenas said. “There are places where we don’t need bipedal robots. My view is that the bipedal form factor will win out, but the question is how do you get out there?”
Not every terrain requires legs. Earlier this week, Diligent Robotics co-founder and CEO Andrea Thomaz revealed that part of the reason her company is targeting health first is the prevalence of ADA (Americans with Disabilities Act) compliant structures. Wherever wheelchairs can go, wheeled robots should be able to follow. Because of that, the beginning does not have to be subject to very difficult problems to build a foot.
But legs have benefits beyond the ability to handle things like ladders. Reaching out is important. Robots with legs are easier to reach the bottom shelf, because they can bend at the legs and hips. You could, in theory, add a huge arm on top of the AMR, but doing so would introduce a whole new set of issues like balance.
Security is something that hasn’t been addressed in the form factor conversation. One of the main selling points of humanoid robots is the ability to enter existing workflows alongside other robotic or human co-workers.
But robots like these are big, heavy and made of metal, so they can be dangerous for human workers. The subject has been top of mind for Wise, in particular, who says more standards are needed to ensure these robots can work safely alongside people.
For my part, I have advocated a more standardized approach to robot demos. Humanoid videos, in particular, have disguised what these robots can and cannot do today. I’d love to see disclosures about playback speed, editing, use of teleops and other tricks of the trade that can be used to trick (intentionally or not) viewers.
“It is very difficult to distinguish what is and is not progress,” said Wise, referring to some new videos of the Tesla Optimus robot. “I think one thing we can do, as a community, is to be more transparent about the methodology used. It adds power to the hype cycle. I think the other problem we have, if we look at what is happening with any humanoid robot in this space, the safety is not clear .There is no e-stop on Optimus. There is no e-stop on most of our robots.
