Humanoid Robots: Engineering the Form Factor

Leo Chen
Will's goal was to be wow. He wanted to create these moments of joy and wonderment for people to experience using these humanoid robots. 

Raymond Yin
Humanoid robots have fascinated us in fiction for over a century, from Maria/Maschinemensch in 1927’s Metropolis, to Robbie the Robot and C3PO, to those more human looking androids such as Data in Star Trek, Ava in Ex Machina, and Megan. Today on The Tech Between Us, we're moving away from the realm of fiction to discuss real life humanoid robots as they enter production. 

To talk about this revolution in humanoid robots, we have with us, Leo Chen, head of US operations for robotics manufacturer, Engineered Arts. We'll discuss the definition of a humanoid robot and dive deep into his company's own humanoid, Ameca. As well as the intricacies in its build and use cases. 

Welcome to the podcast, Leo! Can you tell us a little bit about yourself and Engineered Arts?

Leo Chen
Yeah, my name's Leo Chen. I work for Engineered Arts, which is a humanoid robotics company. We were started 20 years ago by a wonderful man named Will Jackson. Will had this vision of humanoid robots in the world interacting with people. He felt that the superpower of the humanoid form was not in stacking boxes or building cars, but actually was in communication. One of his sort of … I would say famous internally sayings is that the human face is one of the highest bandwidth forms of communication available. And that's definitely true. 

Raymond Yin 
So, he really was way ahead of his time for humanoid robots and especially humanoid robots of that type.

Leo Chen
Yeah, this was something that Will had built from the ground up. He wanted to create these moments of joy and wonderment for people to experience using these humanoid robots. So if you compare our robots with pretty much everything else that's out there right now, we're definitely different. But yeah, I'm lucky enough to run North American operations for Engineered Arts. I've been doing this for the past almost three years now. Before I joined Engineered Arts didn't really have a presence outside of the UK. So, I was tasked with starting this office in the United States, creating this group of wonderful people that now have this unified goal of bringing Ameca, our flagship humanoid robot to as many people as possible.

Raymond Yin 
Like you mentioned, we've had robots for decades, primarily manufacturing, logistics. And I know you guys are often a different direction, but in general, why humanoids and why now? 

Leo Chen
Yeah, robots and manufacturing is not new. In 2024, I think 4.6 million robots were in manufacturing worldwide. And in 2025, we were supposed to add another over half a million robots to that. So I wouldn't be surprised if right now today, in early 2026, we have over five million-ish robots.

Raymond Yin
None of them are humanoid shape, right? Or very few of them are? 

Leo Chen
Well, I wouldn't say none of them. There's definitely pilot programs out there. Agility Robotics, for instance, has partnered with Amazon. They recently signed a deal with Toyota, I believe. And I know that Figure has pilot programs with luxury car manufacturers. And of course, Tesla has their own thing going with the Optimus. But the vast majority of those robots, I'd say over 99% of those robots are not in the humanoid form factor. 

Raymond Yin 
And so in terms of humanoid, is that an optimal form factor for that type of activity?

Leo Chen 
Well, I think it's worth taking a step back and defining what humanoid robots is. So humanoid robot is usually a robot in the humanoid form. So, it has two arms, two legs. It walks utilizing bipedal locomotion. It lifts using one of two arms. And I'd say that we as a society definitely have a fascination with this form factor – especially if you look at popular fiction. 

What are some of the more popular sci-fi movies that involve robots in the past, I don't know, 10, 20 years? You've got Ex Machina, these are humanoid robots, iRobot gets brought up a lot. So popular fiction has us very fascinated with the humanoid form. And the argument you often hear humanoid robot manufacturers put into justifying humanoids in that space is that factories and warehouses are built for humans. To a certain degree, that's true, but I would actually argue that factories and warehouses are built for OSHA compliance and ADA compliance nowadays, right? They're built for safe operation. 

So, within those bounds, you might actually be able to find a better form factor than a humanoid form. OSHA has guidelines around how flat a floor has to be and how clear it has to be. The width of doors is governed or regulated in the United States for ADA compliance. So, what is stopping a creative human being from coming up with a better form factor than us?

And humans aren't very easy to emulate. Humans have incredible endurance. Robots on the other hand though, if you look at the best-in-class humanoid robots right now, I think you're looking at like two to four hours of battery life before you have to swap out. I'm sure there's like quick charging and other stuff or you just rotate your fleet through, but still the endurance isn't quite there. 

The other thing that makes the human form special that we believe at Engineered Arts is the power of communication, right? We can read each other quite well through our facial expressions, through body language and other things. We wanted to build a product that leveraged this superpower of humanity, which is communication. Our ability to create the sense of culture and community with one another through our appearance and how we modulate our face.

Commercial Break: Let's take a quick moment to tell you about Mouser's Technical Resource Library. It's home to articles, videos, ebooks, and more on topics like artificial intelligence, power management, and robotics. Find the information you're looking for by visiting resources.mouser.com.

Raymond Yin
So let's talk a little bit about the direction that you guys are going with the communication side, with the face. And I wish we could do this on video because I would love to show Ameca – kind of the way they look, the way they behave, because it's absolutely extraordinary what you guys have been able to do with all that.

Leo Chen 
I think it reflects on the really intentional design choices that we've made with our engineering team at Engineered Arts and also with the direction that Will Jackson sort of set us on as a founder. We definitely focused on things that are very, very different than other folks. Now, that's not to say that we're not working on bipedal locomotion or that we're not working on dexterity. Our end goal out of all this is still that general purpose humanoid. We still want to get there someday, but I think the path that we're taking is very, very different than other companies. We’re trying to get there by building social products and slowly adding onto the capabilities of those social robots until we reach a general-purpose humanoid versus a lot of these other companies where they kind of have to go from zero to a fully realized product in order to release it properly.

Raymond Yin 
And it's interesting because, some of these other robots, if you look at where the face is, it's just either a blank or a screen or something, whereas you guys have a fully built out humanoid head and are looking to go the other direction with arms and legs and a torso.

Leo Chen 
Yeah, I mean, we have over 30 motors in the head. It's this crazy piece of engineering work. I've got to give mad props to the folks out in the UK, the mechanical design team, the electrical design team being able to fit all of these motors into something the size of a human head and being able to make the face articulate in a way that it captivates people. It draws people in. It makes people want to talk to our robot. 

So right here right now, we're in a couple different applications – entertainment at permanent installations like that is definitely our bread and butter. If you go to the Rosicrucian Egyptian Museum in San Jose, California, you'd see one of our robots dressed up as the Egyptian God, Thoth. Guests can interact with it. They can talk to it while it sort of embodies the character of Thoth, which is really, really cool. That's one of my favorite installations to see because it genuinely looks like a human standing up there talking to you.

Another one is education. Our operating system makes it super easy to code on the robot. Our operating system is called Tritium. Tritium has a web-based IDE. So anybody who has a robot can actually go in with something as simple as a Chromebook and access the code of the robot, change its behavior, and you code in Python for a lot of the behaviors.

But something that not a lot of people think about in terms of education is actually in the classroom as a teaching aid; it's a co-teach with teachers. Having the robot there can enhance the amount of engagement you get with students. So, a lot of these folks come from non-English speaking language families or they might not have a traditional course forward, like they might not go to universities. They might be targeting trade schools – alternative career paths and as a result, they have room to experiment with cool technologies like Ameca.

So, we've got entertainment, we've got education, and those are two core business pillars for us. The final one is receptionist duties. In a high-tech firm, AppDirect, in their lobby, we have an Ameca desktop deployed as a receptionist. In the near future, it'll be linked up into their Slack system, it'll be linked up into their calendar system. We can then agentically focus on things like checking people in, communicating with people, giving people information, that sort of stuff.
Raymond Yin 
Okay, so kind of like an information booth – at any location – engaging with people who walk up to a desk and ask questions and need information, need help, things like that.

Leo Chen 
Yeah, I mean, we support over 50 different languages. It definitely comes in handy to be able to auto translate between languages. If you talk to it in French, it will respond in French. If you talk to it in English, it will respond in English, that sort of stuff. So as a receptionist, it hits a lot of high marks for people.

Raymond Yin 
Yeah. I mean, and I can certainly see that where rather than trying to figure out a kiosk of some kind, like the way we began this, I mean, it's all about communication. How do you communicate? And people want to communicate with somebody that responds in a similar way?

Leo Chen 
Yeah, so our target audience for this type of receptionist role is a company that really wants to make an incredible first impression. Imagine you walk into a lobby and the first thing you see is a robot that can speak your language, that can do all the things that you needed to do, and can draw upon a massive wealth of data. 

Now, we're not looking to replace receptionists or anything like that. Our robots also have a telepresence capability. So, let's say the robot can answer 80 to 85% of the questions or can do 80 to 85% of what a receptionist can do. What this means though is let's say robot hits a dead end in its knowledge base or logic, realizes it can't quite answer things. There's a scenario in the future where it'll be quite easy for it to then phone home to an actual human receptionist that instead of sitting at the front desk, is sitting in the comfort of their own home, they now can talk through the robot, interact directly with a guest, answer the question.

Raymond Yin 
Interesting! Yeah, because I know some of the other human ways like Neo, for example, is almost all telepresence. So that's really interesting that Ameca has that capability, but in a verbal communication way.

Leo Chen
In the near future, we definitely want to expand on these, expand on the capability of the receptionist role, for instance. And we want to dive into some specialty niches for reception work. For instance, patient intake at hospital systems. This is actually one of the highest turnover jobs in hospitals. In some locations in the United States, you're seeing upwards of 25% turnover year over year. So, you're losing a quarter of your staff because of how tough this job is. And we feel like if we can find the correct partner to pilot a program, we can definitely help expand the power of that workforce. We can try to make that job a lot more palatable for the human. So instead of maybe you sitting there at a desk dealing with sick, angry people all day, you can do that from the comfort of your home. Have the robot take care of 80 - 90% of the inquiries, phone in, not have to deal with a sick person in your face angry that they can't schedule an appointment, just sort of look through the robot's eyes instead. 

Other sort of short-term applications in medical that we think we can fulfill are actually patient simulation as well. So, imagine you have a medical school and right now there's a lot of patient simulation with specialized mannequins around things like CPR, surgery, stuff like that.

Raymond Yin 
I've seen those. 

Leo Chen 
But there's also other aspects of being a doctor that aren't really covered by those simulations. For instance, have you ever trained to deal with a staunchly anti-vaccination parent? You probably haven't, but we can set up Ameca to embody that role and have you walk through this with this sort of simulated role.

Raymond Yin 
So instead of developing the medical skills, you're helping a future doctor develop more social skills and more soft skills on working with and engaging with people in addition to being able to do CPR.

Leo Chen
Yeah, 100%. And on the more technical side, the medical side, we can also simulate the face of a stroke patient. Much like in certain types of strokes, you get droop or certain parts of the face freeze. Because we have such fine control over the motors in Ameca’s face, we can help simulate that and basically quiz a prospective doctor or help them recognize the signs of a stroke. So patient simulation is definitely high up on our list of things we want to try to attack short term.

Raymond Yin 
Soft robotics is a fascinating area to me, caregiver bots and whatnot, but that's an area I'd never even thought about where … as a robot that is all about communication, that will give another dimension to potential future doctors and being able to treat real patients. 

So, looking out 10 years, I know especially humanoid is moving so quickly, but where do you see Ameca in 10 years?

Leo Chen 
Yeah, that's a great question. A couple different things that we're looking at in the further term would be companionship robots. One sort of frightening statistic here in the United States is that 94% of our nursing homes here are understaffed. And I think this figure came out three years ago, almost four years ago now. So, I'm pretty sure it hasn't gotten any better. So as someone who is not planning on having kids and has basically resigned himself to ending up in a nursing home someday, that figure terrifies me. And we hope that Ameca, once we've done engineering work, once we've proved it out and validated these hypotheses that we have, can function as a caregiver bot, can provide companionship. 

Some other applications we're looking to in the near future are long-term memory loss care facility care bots. So that 94% figure that I gave was for just general elder care facilities. We dive into specialty care – so things like long-term memory loss, so Alzheimer's, dementia – those sorts of things, the numbers look worse. And one of the things that people who have long-term memory loss need on a pretty regular interval, depending on how far the condition has progressed, is something called reorientation. So, they have to be very gently reminded of, depending on how far the condition has progressed, who they are, where they are, even when.

Raymond Yin 
So, they've got to kind of recenter themselves.

Leo Chen
We think that a humanoid like Ameca can provide the companionship, can provide the general guidance, and really help these people with their quality of lives without maybe putting a burden on a caregiver that's human. 

Raymond Yin
So, when you guys have been developing Ameca, how are you developing its facial expressions and its vocal responses? 

Leo Chen 
Right now, it's actually quite simple. We have set sort of expressions that we blend between the randomness and some other stuff to properly portray the emotion. We're not dynamically generating it yet. That's something that's coming in the near future, but right now it's about creating the sense of connection rather than doing it exactly like a human does. So we focus more on the experience more than emulating exactly how humans do it. And that's actually a core tenet in our engineering. For the face, we're animation first. We're trying to emulate what a human is capable of portraying with their mouth, with their cheeks, with their eyes, their eyebrows, and other things. 

Raymond Yin
So you've not tried to mimic every anatomical feature that governs motion. You guys have just say, okay, to make it look like this, how can we do that? And then when somebody talks to Ameca and the response is scrunching the nose, then it automatically says, “okay, we'll just do this.” 

Leo Chen 
100%.

Raymond Yin
You mentioned that dynamic features were coming. I mean, how would that work? How would a dynamic feature look compared to an animation today?

Leo Chen 
Yeah, so you could basically teach the large language model what a smile should look like. And then when the large language model realizes that, oh, well, this robot should be portraying or expressing happiness or anger, whatever it is, it then sort of dynamically generates that based on motor movements and everything else versus very fixed like, okay, anytime you feel happy, move your lip motors up three degrees, move this da, da, da, da. It's more just more of a natural call like, "Oh, hey, you feel happy." Okay, now pattern match happiness to general facial expressions that portray happiness, how do we do that?

Raymond Yin 
Interesting! Okay, so once again, that could be even coded into the AI models that govern not only the speech, but the motion of Ameca's face down the road.

Leo Chen
Honestly, AI models nowadays, whether they're large language models and then audio models, there's already capability for sentiment analysis, which is looking at a user's text output or speech output and assigning a probability to an emotion. I mean, we do that to the text that Ameca outputs or what Ameca hears. So yeah, it's well within the realm of possibility. 

Raymond Yin
Thanks for joining us on today's episode. We invite you to subscribe and follow for more content including articles, blogs, and more by visiting mouser.com/empowering-innovation. Stay tuned for the next episode as we dive into the realm of quantum computing. We'll explore how these new machines will solve the world's most complex problems, qubit-by-qubit.