The Future of Social Robots: A Step Towards Human-Robot Interaction
Written on
Geminoid F's Uncanny Presence
Geminoid F elicits a unique blend of intrigue and unease. This lifelike android simulates human behavior—breathing, blinking, and speaking in a gentle voice. From a distance, she resembles a young Japanese woman, about 165 cm tall, with long dark hair and soft, smooth skin. Yet, upon closer inspection, her skin is silicon, concealing a urethane foam core and a plastic head mounted on a metal frame. Her movements are facilitated by compressed air, cleverly concealed behind her chair, while her hands rest naturally on her lap. During my visit to the Creative Robotics Lab at the University of New South Wales in Sydney, mechatronics researcher David Silvera-Tawil explained that she was part of a series of experiments.
“For the first few days, I felt a jolt of surprise when entering the room early in the morning,” he shared. “It seemed like someone was watching me. I was aware a robot was present, yet that realization didn’t diminish the sensation!”
The lab director, Mari Velonaki—an artist turned robotics researcher—has been collaborating with Geminoid F's designer, Hiroshi Ishiguro, who is renowned for crafting lifelike androids at Osaka University's Intelligent Robotics Laboratory. Their goal is to explore the concept of ‘presence’—the sensation of another being nearby. They investigate the origins of this feeling and whether it can be replicated in robots.
Even with her extensive experience, Velonaki occasionally finds Geminoid F unsettling. “It’s not repulsive,” she notes, describing an eerie yet amusing sensation. “When the pneumatics are switched off at night, she becomes strangely still…”
This disquiet is linked to a phenomenon known as ‘the uncanny valley’—a term that refers to the discomfort experienced when a robot's human-like appearance doesn't fully match its behavior.
Despite her strikingly human traits, Geminoid F is essentially a sophisticated puppet, valued at around $100,000, driven by algorithms and operated by people behind the scenes. She isn’t designed for intelligence; rather, her purpose is to explore the nuances of human-robot etiquette.
Researchers examining her often come from diverse fields. “We aim to foster collaboration among artists, scientists, and engineers to create robots that can engage with humans in more natural and meaningful ways,” says Silvera-Tawil, now affiliated with Australia's national science agency, CSIRO. The hope is that Geminoid F can facilitate robots transitioning from isolated environments to roles alongside humans, such as caring for the elderly or teaching children.
This transition is expected sooner than anticipated.
The Visionary: Rodney Brooks
Rodney Brooks, often labeled "the bad boy of robotics," has consistently challenged the conventional wisdom of the field. Born in Adelaide, Australia, he relocated to the U.S. in 1977 for his PhD and joined MIT in 1984, where he created insect-like robots capable of traversing difficult terrain with minimal computational resources.
At the time, the prevailing belief was that robotic mobility demanded substantial processing power and sophisticated AI. Brooks, however, reasoned that insects, with their small brains, could navigate effectively. He developed independent "brains" for each leg of his robots, which followed basic commands while a central brain coordinated their movements.
This innovation paved the way for behavior-based robotics, now utilized in various fields, including mining and bomb disposal.
In the 1990s, Brooks pivoted to focus on human-robot interaction, developing humanoid robots that could engage with people and their environments. His creation, Cog, was designed with exposed wiring and mechanical limbs, learning from interactions similar to a child's development. This laid the groundwork for social robotics, where autonomous machines respond to human cues.
Brooks predicts that social robots will soon become commonplace, partly due to demographic shifts—an aging population and a shortage of workers. He points out that the ratio of working-age adults in the U.S. and Europe is set to decline significantly over the next 40 years.
“As caregivers age, there will be fewer individuals available to assist them,” he explains. “We will likely need robots to perform practical tasks, such as carrying groceries upstairs.”
In the 1990s, Brooks co-founded iRobot Corp with two MIT graduates, Colin Angle and Helen Greiner, which produced the Roomba robot vacuum cleaner. This marked the company's entry into the consumer market, with 30 million products sold globally.
Initially, iRobot focused on military robots designed for bomb disposal, known as PackBots. These versatile rovers, equipped with sensors and robotic arms, have been deployed in various emergency scenarios, including the Fukushima nuclear disaster in 2011.
With Roomba's success, iRobot expanded its domestic product line, introducing mops, pool cleaners, and telemedicine robots like RP-VITA, which facilitates remote assessments for hospital patients.
The landscape of robotics has grown, with many companies emerging over the past two decades, producing robots capable of navigating challenging environments and performing complex tasks.
Recent advancements in natural language processing, vision, and machine learning have propelled robot capabilities, allowing machines to handle tasks previously reserved for humans. This creates a feedback loop—enhanced understanding leads to improved performance.
The next frontier involves robots that can interact naturally with people. Brooks emphasizes that advancements in low-cost computing have made real-time 3D sensing feasible. “We can now build robots capable of meaningful interactions with humans, even allowing users without programming knowledge to teach them new tasks.”
In 2008, Brooks and former MIT colleague Ann Whittaker launched Rethink Robotics, which developed Baxter, a collaborative robot designed to work alongside humans without the need for programming. Baxter learns by demonstration, mimicking actions guided by human operators.
Targeting small to medium businesses, Baxter, priced at $22,000, features a digital face with expressive cartoon-like eyes, enhancing its approachable demeanor. It can signal its intentions and even express confusion through its visual display.
Brooks contrasts this with traditional industrial robots, which require precise programming and remain largely unaware of their surroundings. While still vital for tasks like welding and packaging, these robots are typically isolated and not designed for interaction with humans.
Nevertheless, industrial robotics remains a booming market, generating significant revenue through widespread applications.
Brooks envisions a future where collaborative robots—also known as 'cobots'—will extend beyond manufacturing, enhancing various sectors. Baxter gained popularity in research labs, but Rethink Robotics faced financial challenges and ceased operations in 2018, selling its assets to the Hahn Group.
Universal Robots, a key player in this space, was founded by Esben Østergaard, Kasper Støy, and Kristian Kassow, launching the UR5 in 2008. This flexible robotic arm, designed for safe collaboration with humans, has expanded its offerings to include multiple models with increased payload capacities.
While these robots are easy to set up and program, they differ from Baxter in that they do not learn through demonstration but via tablet-based programming. Universal Robots has captured a significant market share, with thousands of units sold and revenues growing substantially.
Brian Scassellati, a former student of Brooks, also foresees robots moving into homes and schools. “We are entering a transformative era reminiscent of the early days of personal computing,” he notes, emphasizing the positive impact on vulnerable populations, including children and the elderly.
Scassellati’s Social Robotics Lab has collaborated with several institutions to develop robots that assist children in learning and overcoming disabilities.
Their goal is to create robots that can adapt to each child's unique needs, guiding them toward long-term educational objectives.
Challenges remain, as these robots must be capable of nuanced interactions, adjusting to each child's individual physical and cognitive differences.
Early results are promising, with studies indicating that interactions with robots can foster educational development. Velonaki’s collaborator, Katsumi Watanabe, observed positive results with autistic children using various robot types, demonstrating improved social skills over time.
Scassellati emphasizes that robots will complement, not replace, teachers. “They won’t take the place of humans but will significantly aid those needing additional support.”
Brooks is optimistic about the rapid advancements in AI and social robotics. He points to challenges faced by technologies like Google’s self-driving car, which initially struggled with real-world scenarios requiring social understanding.
The primary concern surrounding humanoid robots is not their eeriness but the potential impact on employment. Some economists warn that automation may threaten jobs across various sectors, with a significant portion of the workforce at risk.
Brooks argues that fears about job loss due to robots may be exaggerated, citing demographic shifts that will necessitate increased reliance on automated help, especially among aging populations.
He notes, “I’m not concerned about robots taking jobs; I worry about the potential shortage of intelligent robots to assist us.”
Looking ahead, social robots like Geminoid F are unlikely to replace traditional roles but will likely serve as teaching assistants or caregivers. Meanwhile, collaborative robots will increasingly work alongside humans, handling repetitive tasks.
The insights gained from interactions with social robots will inform future designs, making them more user-friendly and less intimidating. Ultimately, they may even become regarded as companions.
THE UNCANNY VALLEY
The term "uncanny valley" was introduced by the Japanese roboticist Masahiro Mori in 1970, describing the phenomenon where humanoid robots elicit both empathy and fear. As robots become increasingly human-like, they can evoke discomfort when they appear almost, but not quite, human.
Mori illustrated this with examples from popular culture, noting that as robot designs approach human likeness, the emotional response can shift from attraction to revulsion.
Real-world applications of the uncanny valley concept have been observed, such as in Pixar's Tin Toy and Sony Pictures' The Polar Express, where overly realistic characters generated negative audience reactions.
Further research has examined this effect, revealing that our brains respond differently to robots depending on their appearance and behavior. If a robot looks human but moves mechanically, it can evoke discomfort due to the mismatch in expectations.
In summary, as robots become more integrated into our lives, understanding the nuances of human-robot interaction will shape their development, possibly leading to more relatable and socially adept machines.