Equipping companion robots with faces and voices inspired by animals or manga style could enhance their emotional appeal, according to a new study from the University of Glasgow. This work could help guide the development of social robotics, a rapidly growing sector.
- Robots with animal or manga-style faces may appeal more emotionally.
- Twenty-four volunteers tested face and voice combinations on the Qoobo robot.
- Faces mattered more than voices, and manga-style faces scored as high as animal ones.
- Emotional readability had little influence, and pet owners reacted differently to the designs.
Social robots can provide companionship and comfort to their owners by mimicking human or animal behaviors. Yet, despite the rapid growth of this technology sector, little academic research has examined how to optimize their design to make them as expressive and emotionally engaging as possible.
Researchers from the University of Glasgow, Shaun Macdonald and Josh Yip from the School of Computing Science, tested dozens of face and voice combinations superimposed on the body of the Qoobo robot, a soft cushion-shaped robot with a wagging tail, with the help of a group of volunteers. They used augmented reality, a technology that adds virtual elements to the view of the real world.
Faces and sounds tested in augmented reality
Dr. Macdonald highlights the value of these robots: “Social robots can offer valuable experiences to a wide range of people. Elderly people have indicated they can help alleviate feelings of loneliness or social isolation.” So-called zoomorphic robots, which mimic the appearance and behavior of animals, are also used by people who cannot keep a real pet. However, he believes that “the aesthetic design of these robots has so far largely relied on educated guesses, with very little research on what makes them appealing.”
The team first created a mood board, a collection of images of animals and robots from real life and popular culture. From this, they derived five prototype faces of animal, robotic, emoji, and Japanese animation-inspired styles, as well as a sixth composed only of eyes. The prototypes also featured five types of sounds: cat meows, vocal sounds close to the human voice, abstract electronic noises, music, and samples of “animalese, ” the meaningless onomatopoeia used by characters in Nintendo’s Animal Crossing game series. The results were published in a paper titled “Anime, Animal or Animal Crossing? Comparing Aesthetic Styles for Zoomorphic Robot Faces and Sounds using Augmented Reality.”
Faces over voices
Twenty-four volunteers, equipped with augmented reality headsets, ranked the different combinations based on the clarity with which they expressed an emotion, the degree of empathy felt, and their perceived relevance. Combinations of animal-style faces and sounds, matching the animal-like body of the Qoobo, were the most favored, judged as more engaging and immersive. However, faces inspired by Japanese animation received equally high ratings, despite not being based on reality.
Josh Yip notes that “manga-style faces are more expressive than animal faces could be, allowing the robot to express itself beyond an ordinary pet, without seeming strange. Study participants still felt they were dealing with a pet-like creature.” Faces consistently performed better than sounds at conveying an emotion and creating a connection. Human-like sounds were easier to interpret emotionally but were less favored than animal sounds.
The researchers were surprised to find that, although participants considered it very important to be able to correctly read the robot’s emotions, this readability had little influence on the preferred designs. The “animalese” sounds, though incomprehensible, were also well-ranked, with some participants believing they suggested greater intelligence in the robot. Finally, pet ownership influenced reactions: pet owners were more likely to treat animal-like prototypes as real animals and to view robotic faces as machines.
For Dr. Macdonald, “no single face or voice is universally favored, but their combination can have a significant effect on how users judge the robots’ ability to communicate emotions.” The work is to be presented at the ACM International Conference on Mobile Human-Computer Interaction in Swansea, Wales, on Thursday, September 3.
