Where Humanoids Really Stand

“Everyone thinks humanoids are 10 years out. They’re not,” said Melonee Wise, chief product officer at Agility Robotics, at the recent Humanoid Robot Forum put on by the Association for Advancing Automation (A3). “They’re here right now. Like right now right now.”

The hype for artificial intelligence (AI) has been extreme as well, but industry is putting generative AI to the test more every day, coming to practical terms with where it can already be useful — and where it still needs some work.

And yet generative AI still can’t take the trash out for you, bemoaned David Reger, CEO and founder of NEURA Robotics. To get it to do the jobs you don’t like to do, give AI a body.

Speaking at the forum, Reger emphasized the smarts going on inside the robot. Beyond that, you need to give your robot the flexibility it needs to get the job done, he said. That’s what makes the humanoid proposition such a valuable one — and one that is a whole lot more than the desire to create machines in our own likeness. Humanoids have the flexibility to work in a world built for humans.

As a third-party logistics provider, GXO Logistics has played a critical role in funding and pushing forward humanoid development. The potential is huge, says Adrian Stoch, GXO’s chief automation officer. The company’s operational incubator program, which currently has 28 active projects, not only provides product feedback but also access to live operations and direct operator feedback — offering realistic logistics spaces for companies to test the ins and outs of their humanoids in action.

But there are still plenty of challenges. Key among them is safety. There are zero cooperatively safe humanoid robots right now, which means they’re confined at the moment to work cells.

Unlike collaboratively safe robots, which are intended for direct interaction with humans, cooperatively safe robots can share a workspace and detect human presence but are generally designed to work with minimal interaction. Autonomous mobile robots (AMRs) typically fall within the cooperatively safe category. As Wise explains it, the human and robot might be working on the same material, but they won’t be touching it at the same time.

Agility Robotics, for one, will have its humanoids out of the cage within the next 18 to 24 months, Wise insisted. “It’s a hard problem, but we’re committed to doing it.”

Safety will ultimately decide whether humanoids have staying power, she added.

With some 72 degrees of freedom and perhaps hundreds of sensors, safety of humanoids is not a simple proposition, according to Nikolai Ensslen, founder and CEO of Synapticon, who discussed the intricacies of functional safety for humanoids.

Take just this one fundamental example: The safe state of a traditional industrial robot is a stopped robot. If you stop a humanoid robot, however, its momentum is likely to continue its forward motion, causing it to fall on the very human you’re trying to protect.

Safety is enough of an issue on the plant floor. Bring humanoids into more public spaces and homes and, “Oh boy, we’ve got some problems,” said Aaron Prather, director of the Robotics & Autonomous Systems Program at ASTM International. “We need more of you working on this.”

And that’s how safety issues are most likely to be solved — by industry working to get there rather than by regulations driving the efforts, according to Brendan Schulman, vice president of policy and government relations at Boston Dynamics. Manufacturers know how damaging it would be for the whole industry if a catastrophic accident were to happen with a humanoid.

Battery life is also a challenge — and could also lead to a safety issue. Just as it would not be safe to put an e-stop on a humanoid, it would also not be safe for the humanoid to run out of battery power and collapse onto a human worker.

But the battery issue goes beyond safety. Most humanoid robots out there today have a battery run time of 2–4 hours, pointed out Ben Waters, CEO and founder of WiBotic. Multiply the frequency of battery swaps needed times the number of humanoids that could ultimately be used by industry, and you’ve got yourself a logistical nightmare if the battery challenge isn’t solved.

You could put a larger battery on the humanoid, but the weight added to the robot would just drain the battery that much faster, Waters added.

He detailed several reasons why wireless charging makes sense for humanoids. Founded in 2015, WiBotic is pioneering autonomous charging for humanoids, AMRs, drones, and other robotic fleets.

Among the list of hurdles to achieve the humanoidability of more jobs, mechanics top the list. Dexterity and delicate manipulation rely on the sense of touch — the most fundamental human sense. It’s the most fundamental humanoid sense, too, argued Joe Michaels, chief revenue officer for HaptX.

“Dexterity and touch control can be powerful tools,” Michaels said, explaining that they will be key to moving humanoid technology forward.

In the exhibit space, HaptX was demonstrating its haptic gloves, which provide haptic feedback to tele-operators in order to train the AI models that govern and manage humanoid robots. Through actuators in the glove, the human hand can feel what the robot hand feels—helping to provide the right touch to perform a task. This technology helps to program complex humanoid robots for standard assembly or material handling tasks, significantly cutting programming time while also generating tons of new data to optimize future developments and applications.

A huge part of the puzzle for humanoid robotics is how to generalize, according to Jeff Cardenas, co-founder and CEO of Apptronik. “There are thousands of robots that do one thing,” he said. “The key is to make one robot that does a thousand things.” But that’s a whole lot easier said than done, he added.

Apptronik is studying the way humans do things to better understand how humanoids might perform those tasks more effectively.

In brownfield applications, “legs win the day over wheels,” Cardenas said. In a facility that wasn’t built with mobile robots in mind, there are too many places that are difficult for AMRs to move around in.

But that doesn’t mean the human form is necessarily the final model for the humanoid. “One of the things you learn building humanoids is how awesome humans are,” said Ritesh Ragavender, founder and CEO of Reflex Robotics. But we don’t have to copy humans completely when developing humanoid robots. “If we’re too stuck on the human form factor, we’re not going to get there.”

Ragavender showed examples of Reflex’s humanoids, which have a somewhat different look than a lot of the other humanoids in development. In some cases, they might even have wheels. The key point is that the application drives the form.

Sometimes, this means having a robot that looks more like a dog than a human. “Depending on the environment, I need the right tool. Maybe a quadruped makes sense,” says Robert Stokes, CEO of Stokes Robotics, an integrator that takes a more agnostic approach to the humanoid space.

Sure, there’s a lot of hype surrounding humanoids. But that’s a good thing, Schulman said, because it provides the attention and excitement the industry needs to push forward. Attention has been lacking for robotics overall from the U.S. government, he noted.

The humanoid industry needs government’s involvement, Prather said, not just for regulatory controls but because they’re the biggest mouthpiece with the most money available to invest.

“This is going to be a real thing because it has to be,” Schulman argued.

Labor issues are pushing robotics and other automation forward throughout industry. “There’s a lot of boring work out there that robots could be doing,” Ragavender says, emphasizing that it doesn’t mean fewer jobs for people. “It’s not a zero-sum world.”

Ragavender’s advice: “Start with one workflow, one floor, one facility. See what humanoids can do.”

Whatever your marketing goals may be, TECH B2B Marketing can help with them all. To learn more about how we can help with your next integrated marketing campaign, contact us today.