Owen Cumming wonders what drives us to make artificial reflections of ourselves. This article originally appeared in the Cosmos Print Magazine, March 2025.
Humanoid robots have been a mainstay in literature and storytelling for centuries. Be it androids, automatons, cyborgs, droids, or golems, we’ve been fascinated by the idea of beings made in our image, replicas of our own bodies and minds.
Talos, the mythical guardian of Crete, is one of the earliest stories we have about a mechanical being created with human form. Forged in bronze by the Greek god Hephaestus, Talos looked like a man, moved like a man, and could even think like a man. While the story dates back to 700 BCE, Talos was – as we would call him today – a robot.
This work of fiction is drawing ever closer to reality. Hiroshi Ishiguro’s Geminoids, Sophia by Hanson Robotics, Optimus by Tesla… the prospect of a humanoid robot that can move, act, and think like a person is on the precipice of being realised. But that raises certain questions.
Do we want humanoid robots in our society, our workplaces, or maybe even our homes? And, if so, why?
Hiroshi Ishiguro and an android made in his image, Geminoid. Credit: ATR Hiroshi Ishiguro Laboratory.
A reflection of ourselves
Associate Professor Janie Busby Grant is psychology lead researcher at the University of Canberra’s Collaborative Robotics Lab. Her research looks at the interactions between humans and robots, and she is keenly interested in our motivations for creating robots with human form.
“There are a number of different factors driving the development of humanoid robots, some of them psychological, and some of them practical,” says Busby Grant.
“Designing a robot that mimics our shape and capabilities means it should be able to travel through doors, stairs and rooms, operate cars and other equipment, and flick switches, turn knobs and operate interfaces the way we do.”
The world around us is designed for human bodies, so, on a practical level, a robot designed to share those environments could benefit from a human form. Of course, there is already a plethora of non-humanoid robots performing specific tasks in human environments – drones, articulated arms, and cleaning robots – but none of them fulfil the wholistic replication of human ability that seems to be a goal of robotics.
There’s more motivating this wave of humanoid innovation than just practicality though. The act of creating something in our own image provides the opportunity for emotional connection and to have those feelings reflected back to us.
“Often designers are trying to encourage people to see the robot as ‘alive’ and to attribute human-like features such as emotions, beliefs and preferences to it. Providing a humanoid body shape makes this easier for us to do and hopefully makes the robot more effective,” says Busby Grant.
Along with the potential practicalities of human-like design and the appeal of emotional connection, there’s another reason we seem to anticipate the development of robots that will move and act like humans. We’ve been told, repeatedly, to expect them.
The humanoid robot Sophia, created by Hanson Robotics. Credit: Anton Gvozdikov / Shutterstock.
“Most people want humanoid robots to look and behave in a certain way, and fiction has been a key factor driving these ideas,” says Busby Grant.
“Companies are catering to social expectations. Movies, TV shows and fiction have all told us that robots will look roughly like us, and that technology will deliver us humanoid robots sooner rather than later.”
There’s the additional narrative, bolstered by an epidemic of loneliness, that robots will be sophisticated enough to act as surrogates for human companionship, or even intimacy. A society where a human persona can be substituted by a robot could face serious ramifications, but this, at least, may not be an immediate concern.
“There are serious social and ethical concerns around designing robots to replace or augment human companionship. But the systems we have currently, and in the near future, are not going to be able to replicate the complexity, nuance and positive outcomes of human engagement,” says Busby Grant.
It’s hard to say exactly what roles humanoid robots may play. Mild mannered butlers serving at our whims, intimate companions for a lonely population, or perhaps indominable mechatronic conquerors of humanity. Who’s to say.
Humanoid robotics is fuelled by a dream to embody our ideals and fulfil our expectations, but a dream only takes you so far. Eventually, someone must work out how to actually do it, which, in this case, means building some of the most advanced machines to have ever existed.
Layers of technical complexity
“There’s not one day that I go without thinking how amazing our body is,” says Professor of Robotics and Art, Damith Herath, who founded and leads the Collaborative robotics lab at the University of Canberra.
For someone attempting to recreate the human form, every intricate detail is illuminated. Every sense, every structure, and every movement of the human body becomes an engineering marvel. But with decades of experience designing and building automated robots, Herath is painfully aware of how difficult it can be to create robots that look and move like their creators.
“We are trying to create humanoid forms that can work in the same environments as humans, but humans are complex beings,” says Herath.
“For example, there are about 7 degrees of freedom just within the human arm. To expand that to some sort of humanoid form, you’re talking about hundreds of degrees of freedom.”
One degree of freedom is the ability to move one joint in one direction, like bending and unbending your elbow. So, with somewhere between 250 and 350 joints in the human body, there’s a lot for roboticists to consider.
“All of these different complexities need to work in tandem. You’re looking at hundreds of motors and actuators, each one monitored through a central system. There are a huge number of different algorithms working to just keep the whole system from falling over,” says Herath.
“And that’s without achieving any specific goal. If you want to get something done, that adds layers of mechanical, algorithmic, and societal complexities as well.”
Even for tasks that humans might consider mundane, the complexity of the programming and engineering involved is baffling. Two PhD students on Herath’s team are trying to ‘grasp’ how the human hand performs a task as simple as sorting rubbish.
“People can easily identify textiles or organic materials and pick them up without even thinking, right? But a plastic bag, versus a rotten fruit, versus a T-shirt – each requires a different kind of dexterity. We still don’t have a robot that can actually do all of that in one go. It is such a such a complex problem,” says Herath.
But inherently, the human form implies that these robots are designed to work with and alongside humans. So, even once the mechanics have been worked out, what really makes a humanoid robot successful is their ability to interact with people.
Professor of Robotics and Art, Damith Herath, who founded and leads the Collaborative Robotics Lab at the University of Canberra. Credit: Liam Budge.
The human touch
Professor Mari Velonaki is the director of the Creative Robotics Lab at the University of New South Wales, a research group specifically focused on the integration of robots into human society.
“Our aim is to produce technologies that enhance human experience,” says Velonaki.
With the rapid advancement of humanoid robotics, the Creative Robotics Lab helps to evaluate new technologies and design robots that are right for social implementation. When it comes to designing humanoid robots, one of the major challenges that Velonaki’s team deal with is deciding how ‘human’ a robot needs to be.
“When we say humanoid robot, it’s a very big category. The human references might be quite abstract, but you still can see something that appears like a head or the camera is like eyes. We can go from something very near realistic to something a bit more abstract. It’s all humanoid robots,” says Velonaki.
Gauging what features a robot should have requires a deft touch. Human features can help us feel an affinity for machines, but some of the most lifelike humanoid robots, like Hiroshi Ishiguro’s Geminoids, can become victims of the ‘uncanny valley’, the dreaded sensation of wrongness that comes from a robot seeming almost human, but not quite right.
“It’s something that starts with attraction and turns to some sort of repulsion,” says Velonaki.
Making humanoid robots fit for human society isn’t simply a matter of creating human replicas. Rather, the implementation of humanoid robots is about understanding what elements of humanity a robot needs.
Professor Mari Velonaki is the director of the Creative Robotics Lab at the University of New South Wales. Credit: Courtesy of Professor Mari Velonaki.
“We need to think, ‘why does this robot need to be humanoid?’, and design for the situational context from the very beginning. Where is it going to be operating? What will it be doing? Who is interacting with it?,” says Velonaki.
The potential ways for humans and robots to interact are boundless, but it’s not one-size-fits-all.
“Small, toy-sized humanoid robots might be helpful for young children on the spectrum to learn about social cues and social behaviours by playing. That’s a good match. But for people in a nursing home, something that looks too much like a human can be highly problematic,” says Velonaki.
“Aging populations can have sensory systems that might not be fully functional, and it creates an area of confusion, ‘Is that a human or not? In that context, you need something that doesn’t look threatening, and has some sort of human references, but is very clearly a robot.”
Ultimately, humanoid robotics is a science of interaction. Function is good, the robot needs to do what we want it to do. Trust is even better, something that we can look at with surety and comfort. Yet regardless of how well they might function, or how friendly they might look, for humanoid robots to really work in a human world, they need to have something that we want to interact with.
“There’s nothing wrong with making something function as a matter of fact, but why don’t we inject creativeness and playfulness into the objects or technologies that we use every day?” says Velonaki.
Humans are playful; it’s in our nature. Playfulness is part of how we converse with the people and things that we care about, and creating that sensation can begin without even the need for language. When we speak to another person, perhaps 70% of everything that passes between us is non-verbal. This tapestry of our expression is part of our humanity, informing others of our thoughts, moods, and intentions.
“If you’re whispering, if you’re singing, if you’re moving, people know what you mean. These things are all cross-cultural and this is a great foundation that we can start building on,” says Velonaki.
We’re hardwired to understand human expression, and even simple gestures can hold a depth of meaning. So, when we’re designing ways to communicate with robots, why not use the signals that our brains already know?
“It’s very important for emotional and mental health to be a little bit more playful with these technologies. An expression of a face is better than written text in terms of long-term cognitive load fatigue. It’s much faster and easier to understand,” says Velonaki.
“If it’s just a screen with a big smile, I don’t have to go close, I don’t have to put my glasses on, I don’t have to read more text, I don’t have to think about it!”
From Velonaki’s perspective, the purpose of humanoid robots entering society isn’t to create more work for people, or to replace the relationships that we already have. It is to make our lives easier and more enjoyable, taking on the jobs that we don’t want so we can focus on what matters most.
“We want to be able to have humans do what humans do best, and that’s connecting with people,” says Velonaki.
“It’s holding Mrs Smith’s hand and explaining to her what’s going to happen before surgery, rather than destroying my lower back because I have to carry Mrs Smith from the bed to the wheelchair. I can see a robot in between, helping and making it entertaining, but always with a human in the loop.”
Guiding principles
Humanoid robots are only going to become more advanced as our technology develops. We need to prepare ourselves for radical changes in our societies, cultures and systems of commerce.
“Once AI is embodied in actual mechanical systems and they start to move around in the environment, that’s going to disrupt the whole host of industries and socioeconomic stratums,” says Herath.
Roboticists and engineers are increasingly aware that their passionate drive for creation needs to be tempered by those who can provide perspective on how humanoid robots may affect society. As Herath puts it, quoting Professor of Electrical Engineering and Computer Sciences, Jitendra Malik, “Robotics is too important to be left in the hands of roboticists.” Roboticists need to work alongside sociologists, psychologists, and ethicists to understand and scrutinise their potential impact.
“We need to be both realistic about what robots can do but also have the debates in coming decades about what we as society want them to do,” says Busby Grant.
The complexities of having robots with a human form, and some semblance of a human mind, entering our society won’t be quickly resolved. We will need to reflect upon our own morals and motivations. How should a machine that is humanoid, but not human, be treated? Are they a practical solution to our problems, or just a fancy of fiction we’ve brought to reality, creating more problems than they solve?
A guiding principle when designing a technology that imitates humans, could be to create something that brings more humanity into the world, rather than just more technology.
“Well designed robots don’t make us like machines, they enhance our humanity. We want to design robots that remind us we are human,” says Velonaki.