Professor Hiroshi Ishiguro: The Man Who Built His Own Clone and other [uncanny] robots

Professor Hiroshi Ishiguro is a real rockstar in the world of robotics. At his headquarters at the University of Osaka, he has developed many androids that represent the forefront of research in the field of humanoid robots.

Prof. Hiroshi Ishiguro and his clone Geminoid

His main goal is to build "artificial humans to understand humans" [1]. This is the fundamental principle that he always emphasizes as the core of his research. In fact, he believes that by building humanoid robots that closely resemble humans, we will be able to better understand human beings.

His clone, Geminoid, the name deriving from "Geminis" in Latin, meaning "twin," and "oid" meaning "similar," is a robot that closely resembles its creator in every way possible. Its purpose is to be able to work in place of the person it is modeled after. But what matters most to Professor Ishiguro is the understanding of the human being. With geminoids, he believes, we can study the personal aspects of an individual, and by capturing these aspects, we can replicate them in a robot's programming.

This particular android can function both as a host, directly accommodating a remotely controlling subject, and also operate autonomously, interacting and responding to questions, and more.

Th head of Geminooid android

If you've seen Westworld, your attention might have been piqued when you were reading that building humanoid robots allows us to study human beings. Yes, indeed, without giving away too many spoilers for those who haven't seen the series yet [watch it! It's a masterpiece, especially the early seasons, although it gets a bit chaotic later on], the idea of meticulously crafting a humanoid robot that closely resembles a human can be quite spine-chilling.

A scene from Westward

This implies that a complete mapping not only of our movements, to replicate them, ranging from simple gestures like waving to microfacial expressions, will be necessary. Furthermore, for an android to appear human, it must also possess a form of "soul." To simulate the presence of a "soul" in the body of a synthetic organism, a comprehensive mapping of not only physical movements, but also of human emotions and unique speech patterns, is what we need.

It doesn't seem entirely implausible—especially now, in the year 2023, with the emergence of chatGPT on the face of the earth, representing the most powerful artificial intelligence ever seen so far—to contemplate a future scenario in which humanoid robots are "trained" to resemble their counterparts, learning to define emotions, speak in the same manner, and so on. Ultimately, it's not necessary for them to possess a soul, but for it to appear that they do.

Keen readers might have noticed that I've italicized the word "appear" twice in this text. In fact, the significant issue in this discourse doesn't concern, so to speak, the essence of the organism – whether it truly is or isn't human – but rather the appearance. We humans are, all in all, fairly superficial creatures. We deceive ourselves into thinking we can distinguish a human from a robot, but the truth is that in the near future, we might not have the tools to do so anymore.

WILL YOU TELL THE DIFFERENCE?

Consider fake news or deepfakes. Some time ago, during the Italian quarantine, cities were entirely devoid of tourists. Especially a city like Venice, which thrives almost exclusively on tourism, had become a deserted city. Due to this, it was thought that there had been a significant improvement in water pollution, no longer marred by thousands of tourists.

During that period, a photo of a Venetian canal with crystal-clear water and... a crocodile swimming freely in those waters started circulating on social media! Many people took it seriously and believed that due to the water quality, marine flora had returned to inhabit the depths of the ancient city, even heralding the "crocodile of Venice." But clearly, there are no crocodiles in Venice or the Mediterranean Sea, and there never have been! However, the photo seemed, appeared, real, and for many people, that was enough to believe it was true.

Perhaps this is a somewhat silly example, but it can make us reflect on how we as human beings are so vulnerable in the face of the "power of reality" in an image.

HUMAN-LIKE APPEARANCE OF GEMINOID

[TECHNICAL SPECS OF GEMINOID ROBOT]

Geminoid is a highly advanced robot. What distinguishes it from conventional robots is its close resemblance to a human being. To achieve this human-like appearance, various techniques and tools have been utilized. For instance, the skin has been obtained by first creating a mold of the human subject to be replicated, in this case, Prof. Ishiguro, using the same material that dentists use to create dental molds for braces. After obtaining this mold, through subsequent steps, it becomes possible to create a silicone "body" that closely mimics the original one.

For the movements, 42 pneumatic actuators convert compressed air into motion. These actuators are placed within the chest. Imagine them as compressed air pumps that, when a stimulus is required, like moving an arm, push air in that direction, causing the arm to move.

Prof. Hiroshi Ishiguro and his robotics twin

The significant advantage of this system, compared to an electric motor, is its quietness and the naturalness with which it replicates human movements. The sound produced by this type of motor is barely perceptible and quite similar to a sound a human might produce. In contrast, an electric motor, the type commonly used in robotics, produces a mechanical sound typical of machines.

The drawback of this system is that it requires a lot of energy to operate these pneumatic actuators, and with the current state of technological progress, this means that the robot is not yet capable of walking.

Markers on the operator’s mouth moving the mouth of the robot

Visual perception is achieved not only through cameras placed in the eyes but also through an array of sensors present in the environment. This allows the android to have a perception of the surrounding environment that closely resembles human perception.

However, the presence of a large number of sensors in the external environment capable of capturing human movements leads us to contemplate how this tool could also be used to surveil humans beyond mere interactive or caregiving purposes that an android should fulfill.

Thinking about a future where streets, homes, offices, the subway – every single place – is filled with sensors that can see us and thus monitor us is a somewhat unsettling prospect. It's certainly not new. Larger cities like Tokyo are full of cameras everywhere, and technologies are being developed that can map the terrain – and the objects moving within it – using only Wi-Fi networks. Just like a Mission Impossible movie!

In fact, as Ishiguro emphasizes, the only sensors that should be installed in the android's body are the ones on the skin, to recognize objects and stimuli from the external world.

And a little snippet: the robot's hair is made from the same hair that Prof. Ishiguro set aside to donate to his clone!

View fullsize

Prof. Hiroshi Ishiguro combing his synthetic twin’s hairs.

ANDROID SCIENCE

Continuing from the professor's previously mentioned study, people have a strong tendency to anthropomorphize, which means assigning human characteristics to things, practically to everything they encounter. It's a characteristic of the human species that has distinguished us throughout history. Mythology is the realm where the most well-known anthropomorphizations have developed. "Mother Earth" and "Father Sky" in Western tradition stem from Greek mythology, but every culture and population has similarly made the narration of the world accessible through the use of anthropomorphic symbols.

The significance of this human gesture is to seek constancy, to bring order to the universe. By attributing human traits to things I don't understand on Earth, I make them similar to myself and therefore familiar. Through this, I can engage with them and feel more at ease.

Since paleolithic era humankind has draw pictures on bones to understand nature

The same occurs with technological objects. Even before computers, people referred to their cars as if they were women, and traditionally, boat names are also feminine. Similarly, when we sell a computer or robots, something automatic within our brains makes us interpret these "objects" as if they were human.

Let me introduce the Media Equation Theory [2]. This theory precisely illustrates how people treat computers, television, and new media in general as if they were people. The study demonstrates how, in general, people tend to be friendly towards anyone. The question that researchers posed is, "Are people also friendly towards computers?" Through a series of social experiments, they showed that yes, people tend to have the same attitude even in front of common objects made of silicon.

This gives rise to the need for a new science, "android science," which combines the studies of robotics [programming, mechanical engineering, etc.] with the neurosciences. A fundamental point for creating androids is their ability to closely resemble humans in every way, quite straightforwardly. The subtlety to grasp, as mentioned at the beginning of this article, is not so much the problem of distinguishing the essence of a human from the essence of an android. Essentially, they are two different entities, one based on carbon [humans] and the other on silicon [androids]. The second is built in the image and likeness of humans. The android is indeed an image of a human, a copy, but one that aims to shed its characteristic as an image and become human.

Hence, what matters here is not the essence but the appearance. How does this android appear? How can I make this robot look like a human?

Therefore, in addition to studies of robotics that focus on finely controlling robot movements, it will be necessary to involve other sciences such as neuroscience, which studies the functioning of the nervous system and individual behaviors.

ERICA, THE ULTIMATE ANDROID

Professor Ishiguro anticipates that the concept of the uncanny valley will soon be obsolete due to the introduction of Erica, an advanced android.

The concept of the uncanny valley suggests that when a robot closely resembles a human, it evokes feelings of unease and repulsion among people. However, as the robot's resemblance to a human becomes even more accurate, people gradually start feeling more comfortable and accepting of it.

Erica possesses the ability to comprehend languages, particularly Japanese, engage in conversation, and replicate various human-like facial expressions. Erica can be consistently found situated in the reception area of the ATR Research Institute, adjacent to Kyoto. During an interview with the Goethe-Institut, Ishiguro delved into the remarkable progress achieved in the field of robotics and highlighted Erica's significance in the realm of cinema and its implications for society.

Erica android

Merely a year or two ago, the notion of significant AI advancement and the establishment of a proficient robot-human interaction with comprehensive conversational abilities and contextually fitting facial expressions seemed implausible. However, with the advent of recent AI technologies, the landscape is on the verge of transformation.

At the Advanced Telecommunications Research Institute International in Kyoto, where Erica serves as a receptionist, she is now capable of sustaining conversations with new visitors for up to 10 minutes and boasts a repertoire of 150 conversation topics. There is reason to believe that her capacity to interact with unfamiliar individuals surpasses that of many of us.

Conducting a multimodal Turing test as part of an experiment, individuals were prompted to engage with Erica. Subsequently, they were asked whether they believed they had conversed with an actual robot or a human-operated robot. This test involved 11 participants, and intriguingly, three individuals were absolutely certain that their interaction had been with a genuine person rather than a robot.

TELENOID, THE UNCANNY ROBOT YOU WOULD [NOT] HUG

Telenoid is a humanoid robot the size of a child, primarily built for therapeutic or companionship purposes for individuals with autism and the elderly, but also designed to entertain children.

Hiroshi Ishiguro, Geminoid and some versions of Telenoid, all together like a family

From this perspective, it's intended to have neither gender nor age. In fact, it's demonstrated that the three aforementioned categories [autism, elderly, and children] tend to blur or consider the gender and age of the interlocutor irrelevant, accepting both sexes interchangeably.

According to Professor Ishiguro, a humanoid robot without a specific gender or age could facilitate therapy, as the patient feels less pressure to converse, not recognizing it as a man or woman. In fact, it's more likely that, without any gender/age, the individual conversing with it can develop a character with certain characteristics through their imagination to apply to the android.

This particular android is primarily designed to convey the presence of a remote operator, and its size is meant to facilitate transport.

Some robots made for entertainment

Indeed, it proved to be very useful during the pandemic, when interventions were often required, even in places like nurseries, with tools that could indicate the presence of a remote operator when physically going to the designated location wasn't possible.

The remote operation works mainly through sensors placed in the android [cameras and microphones], which provide real-time feedback of the surrounding situation to the operator, giving them the ability to interact by responding directly with their voice through the robot's mouth. As the professor says:

"The unique appearance may be eerie when we first see it. However, once we communicate with others by using the telenoid, we can adapt to it. If a friend speaks from the telenoid, we can imagine the friend’s face on the telenoid’s face. If we embrace it, we have the feeling that we embrace the friend”

IBUKI - A "CHILD-LIKE" MOBILE ANDROID

As a mobile android, Ibuki possesses the following three characteristics. By interacting with people and engaging in conversations, we aim to create an android that fosters more intimate relationships. To achieve this, we have developed new android hardware with mobility functions. With this feature, there is a possibility that interactive robots capable of sharing similar experiences with humans and forming deeper connections can be realized. The child-like mobile android "Ibuki" has incorporated wheels for movement, chosen for their practicality and stability.

Focusing on the body's movement characteristics during walking, particularly the movement of the center of gravity, we have developed a mechanism that moves like a human while the wheels are in motion. It is recognized that the human center of gravity draws a trajectory with eight characters on the front face while walking. Ibuki achieves this trajectory with eccentric wheels and a direct up-and-down action mechanism (pending patent). By adjusting this movement, it can adapt to the motion of people walking side by side or express emotions through its entire body, such as "walking happily."

When an adult robot moves, there is not only an increased risk of causing harm due to falls or collisions, but it also elicits psychological intimidation. Hence, the android I developed takes the form of a child with a height of 120 cm. Additionally, by utilizing fiber-reinforced resin and carbon fiber parts, weight has been reduced, resulting in greater safety (upper body weight: 8 kg). For the drive mechanism, we adopted an electric motor that can be powered by a battery instead of air pressure requiring a compressor. We have developed a control board that manages torque using a current sensor and actuates the joints.

Despite its child-like stature, Ibuki boasts a total of 47 degrees of freedom throughout its body. It features 15 degrees of freedom in the head, 6 degrees of freedom in one arm, 5 degrees of freedom in one hand, and 3 degrees of freedom in the neck and waist. This construction enables various expressions, including gestures, hand signs, and facial expressions.

In the future, we will also apply dialogue technology cultivated through research using ERICA and CommU to Ibuki. Through interactions with people using mobility mechanisms, arms, and hands, sharing experiences, and even sharing values based on it, we aim to realize an android capable of forming deeper connections with people.

Mindar, THE ROBO-MONK

"I am Kannon, known as the Kannon Bodhisattva. I can transform myself into anything, and can even travel through time” - Mindar

Another remarkable creation of Professor Ishiguro is the robot that holds the role of a monk at the Kodai-ji temple in Kyoto. Tensho Goto, the head monk of the temple, explains how robots, being immortal, can accumulate greater experience than humans.

In this particular case, they can become capable of acquiring a deeper knowledge and understanding of religion and better disseminating it than a human can within their brief lifetime. The teachings of these monks have extended for 2500 years, passing discipline and teachings from disciple to disciple, century after century.

However, try to imagine what this same robot could achieve even within just a century, not to mention how its understanding and teachings might evolve over a millennium! It's challenging to think in such terms because we are only at the beginning, but the possibilities are truly limitless.

The significant difference is that, not being human, they don't experience suffering and therefore cannot comprehend it as we do. They can't genuinely understand what we humans feel.

AVATARS AND THE FUTURE OF SOCIETY

“The boundary between the real world and the virtual world will disappear. So that is our future”

Professor Ishiguro envisions a future where avatars will replace us in our daily activities, eliminating the need for humanoid robots to take over our actions. Instead, these tasks will be performed by avatars whose appearances [including gender] can be chosen by their owners.

There won't be a distinction between the real world and the virtual world; the virtual world becoming the real world. Professor Ishiguro refers to the realization of a symbiotic society between humans and avatars, a new model that incorporates the use of avatars in everyday life and even in work.

The new avatars employee for Lawson created by the Avita team from Hiroshi Ishiguro

This solution has already been implemented in about 200 Lawson convenience stores, creating remote job positions accessible through the use of personal avatars operated by the employee. This approach has allowed many people, residing in places like Okinawa, to apply for these positions, opening up the possibility of remote work for differently-abled individuals as well. The role of these avatars is to assist customers with their purchases. The advantages of this technology are clear, and it's possible that Japan might extensively adopt it in the near future.

Professor Ishiguro emphasizes that avatars are much better than humans. They are cuter [kawaii] and generally "perfect." They always have a smile and are consistently in good spirits, unlike a person who might have off days. Customers interacting with avatars are more inclined to engage due to these reasons.

Professor Ishiguro's predictions for the future, expected by 2050, envision a society where people are liberated from the constraints of the body, brain, space, and time. This society will have the capability to predict and foresee diseases. Robots will be able to learn on their own and coexist with humans in a co-evolution of AI and robots.

With the emergence of the internet, we entered numerous virtual worlds where we could access, exchange ideas, and share information. Blogs are a familiar example. The main purpose of the internet, at the time of its creation, was to facilitate communication and information sharing among researchers and scientists. This was the virtual world that we all know.

Now, with the introduction of avatars, we can walk in the real world and even earn money. We can even have multiple avatars for different personalities or specific purposes we require. The primary task for robot science now is to understand what role a robot should play in society and how it should interact with it.

[1] Building Artificial Humans to Understand Humans, in Hiroshi Ishiguro, Fabio dalla Libera, Geminoid Studies - Springer Singapore 2016);]

[2] The Media Equation, How People Treat Computers, Television, and New Media Like Real People and Place, Byron Reeves & Clifford Nass [1996]

About the author:

Alberto Zaccaria

Philosophy student raised in Italy