humanoid4rent
登录获取报价

How audiences react to humanoid robots: what the research on public settings shows

Robots draw attention easily, but studies show only a minority of passers-by stop, curiosity fades, and the uncanny valley is less consistent than its fame suggests.

作者: humanoid4rent 编辑团队 · 发布于 2026年11月13日

本页面尚未翻译
您看到的是英文原文,因为目前还没有中文版本。没有任何内容被省略:这是完整的页面,以其撰写时使用的语言呈现。

Peer-reviewed research gives event teams a mixed but usable picture. Robots reliably catch people's attention, yet in field studies only a minority of passers-by actually stop. Curiosity tends to fade as an encounter goes on, and the much-cited uncanny valley shows up less consistently in experiments than its fame suggests. In practice, speech problems and crowding have caused more visible trouble than the robot's appearance.

Why the evidence needs careful reading

Researchers have published little about humanoid robots at live events specifically. Most studies of how people react to robots run in laboratories, with small groups of volunteers and short sessions. The field studies that do exist mostly took place in shopping centres and museums, often with small or wheeled robots rather than the full-size walking humanoids now offered for hire.

The studies below were chosen because they either took place in real public spaces or pull together many experiments. Each one comes with limits, which are stated alongside the findings. None of them measured a trade-show booth or a gala dinner directly, so applying them to events means drawing an analogy, not quoting a measured result.

The uncanny valley: a powerful idea with uneven evidence

The term comes from a 1970 essay by the Japanese roboticist Masahiro Mori. An English translation authorised and reviewed by Mori was published in 2012 and is reproduced by IEEE Spectrum. Mori proposed that people warm to machines as they look more human, until something appears almost but not quite human. At that point, he argued, affinity drops sharply into unease. He also suggested that movement deepens the effect. His advice to designers was to aim for a moderate degree of human likeness instead of near-perfect realism.

Some later experiments support the idea. In a 2016 study in the journal Cognition, Maya Mathur and David Reichling showed participants 80 real robot faces. They measured how likeable the faces seemed and, through an investment game, how far people would trust them. According to the paper's abstract, the valley pattern appeared in likeability and reached even further into the implicit trust decisions.

The overall evidence is less tidy. A 2025 systematic review in Frontiers in Psychology by Ștefania Cihodaru-Ștefanache and Ioana Podina examined 29 studies involving 4,153 users. Only 8 of the 29 found significant support for the effect, and the authors rated most of the studies as weak to moderate in methodological quality. That review covered on-screen conversational characters, not physical robots, so it cannot be applied directly to a humanoid on a show floor. What it does show is that the effect is not a fixed rule.

For event teams, the practical reading is cautious. A very lifelike face that moves imperfectly carries the risk Mori described. Most humanoids hired for events have visibly mechanical heads or screen faces, and are some way from that near-human zone. However, the research cannot say exactly where any particular model sits on the curve.

Attention is easy; getting people to stop is harder

One of the largest field studies with a humanoid form comes from Japan. Yuki Okafuji and colleagues, writing in the International Journal of Social Robotics in 2022, placed Sota, a table-top humanoid about 0.3 m tall, in a shopping-mall corridor. The robot ran for three days in July and August 2019, six hours a day, and tried to get passers-by to stop and take a flyer.

The researchers compared three behaviours. With a simple greeting, about 6.1% of the more than 26,000 counted passers-by stopped. When the robot danced, about 8.6% stopped, but people did not listen to it for long. When it acted as if it were "in trouble", muttering and asking for help, the stopping rate rose to about 10.6%, and this behaviour also did best at getting flyers taken. The authors reported that the robot performed comparably to human advertisers they tested later in the same mall.

Two points matter for brand teams. First, even the best behaviour stopped roughly one passer-by in ten. Second, how the robot behaves changed the result more than the simple fact that a robot was there. The study has its own limits: it used a small robot rather than a full-size humanoid, and the tracking system could not reliably detect small children.

Novelty: the effect that flatters short events

Researchers in human–robot interaction use the term novelty effect for the extra interest people show in something unfamiliar, which tends to fade with repeated exposure. In a 2022 perspective article in Frontiers in Robotics and AI, Catharina Smedegaard argues that the field has too often treated novelty as background noise to be removed before measuring anything. She suggests social robots are novel in a deeper sense, because they unsettle everyday categories such as alive and not alive. On her reading, many reactions recorded today may reflect novelty rather than stable preferences.

A museum trial in Italy shows a version of this within a single visit. Researchers from the Istituto Italiano di Tecnologia and Ericsson deployed R1, a wheeled humanoid guide, at two Turin museums for a week each. Their 2024 paper reports good engagement overall, based on more than 100 questionnaires. Engagement was highest at the start of each tour, and some visitors drifted off to explore by themselves as the tour went on.

For a one-day launch, novelty works in the organiser's favour. It becomes a trap when early excitement is used to justify a long installation or a repeat booking, because the same audience may react very differently the second time.

What happens around the robot: crowds, children and space

The Turin study also recorded practical frictions. The researchers say children were especially engaged and often crowded around the robot more closely than adults did. The biggest complaint from visitors was about speech: replies were sometimes slow or missing, partly because of background noise, the robot's own motor sounds and the difficulty of telling which person in a group was speaking. At busy times, network congestion made the robot stop.

How people move around robots has also been measured at scale. A Science Advances paper published in August 2026 by Dominik Wojcikiewicz, Aude Billard and Diego Paez-Granados analysed four public datasets from Switzerland, the United States and South Korea. The datasets cover tens of thousands of pedestrians passing wheeled robots and a Spot robot dog. Pedestrians left more space around faster robots than around other people, and the gap grew roughly in line with robot speed. In denser crowds, that extra space largely disappeared, which the authors attribute to people having no room to keep their distance.

None of the robots in those datasets was a humanoid, so the numbers do not transfer directly. The broader pattern still matters for planning: in a packed hall, visitors will end up close to the robot whether they intend to or not.

Public attitudes: the background mood

Survey data on general attitudes is older than the field studies. In the European Commission's Eurobarometer on digitisation and automation, published in May 2017, 61% of respondents held a generally positive view of robots and artificial intelligence. That figure rose to 75% among people who had heard or read about AI in the previous year, compared with 49% among those who had not. The same survey found 74% expected robots and AI to destroy more jobs than they create.

These figures are nine years old and come from before today's generative AI tools. Attitudes may well have changed since. The sources reviewed for this article did not include a more recent EU-wide survey asking the same questions about robots. Even so, the 2017 results point to a mix that event teams still meet: general curiosity, alongside some unease about what robots mean for work.

What this means for event organisers

  • Plan for a minority to stop, and design for them. Field data suggests most passers-by will look and walk on. Set engagement targets on that basis and decide what a stopping visitor should do next. The guide to measuring the return of a robot at an event covers what to count.
  • Behaviour matters more than presence. Scripted interactions that invite a response did better than generic greetings in the mall study. Agree the interaction design with the supplier early, as described in using a humanoid robot at a trade-show booth.
  • Treat speech as the weak point. Noise and group conversations caused the most visible failures in the museum trial. Check what to expect in speech, languages and noisy venues, and have a fallback such as a screen or a human host.
  • Expect children and crowds to come close. Plan barriers, operator positions and stopping procedures with the safety planning guide for public events.
  • Do not judge a long run by day one. Novelty inflates first impressions. If you are considering a longer deployment, measure on later days as well, and compare form factors using humanoid, robot dog or telepresence.

Frequently asked questions

Is the uncanny valley a proven effect?

Not consistently. Masahiro Mori proposed it in 1970, and some experiments, such as a 2016 study of 80 robot faces in Cognition, found the pattern in likeability and trust. A 2025 systematic review of 29 studies of on-screen conversational characters found significant support in only 8, and rated most studies as weak to moderate in quality. The effect appears in some conditions but should not be treated as a fixed rule.

How many visitors actually stop to interact with a robot?

Usually a minority. In a 2022 field study in a Japanese shopping mall, a small humanoid robot stopped about 6% of counted passers-by with a greeting, about 9% with a dance and about 11% when it acted as if it needed help. These figures come from one robot in one mall, so they are a reference point for planning rather than a forecast for any particular event.

What is the novelty effect in human–robot interaction?

It is the extra interest people show in an unfamiliar robot, which tends to fade with repeated or longer exposure. Researchers have often treated it as noise in their data, though some argue it shapes how people make sense of robots far more than assumed. For events, it means first-day enthusiasm can overstate how a robot will perform over a long installation or with returning visitors.

Do children react differently to robots than adults?

Field reports suggest they often engage more and get physically closer. In a 2024 study of a humanoid guide robot in two Turin museums, researchers noted that children were particularly engaged and often surrounded the robot more closely than adult visitors. This is one reason event safety plans usually include barriers, a trained operator nearby and a clear way to stop the robot quickly.

How this article was made: written by the humanoid4rent editorial team with AI assistance; every factual statement was checked against the sources listed below on 2026-11-13. Spotted an error? Write to info@humanoid4rent.com.

资料来源

本文中的事实已在撰写当天对照以下页面进行核实。

  1. The Uncanny Valley: The Original Essay by Masahiro Mori – IEEE Spectrum
  2. Navigating a social world with robot partners: A quantitative cartography of the Uncanny Valley (Cognition, 2016) – Europe PMC
  3. The uncanny valley effect in embodied conversational agents: a critical systematic review of attractiveness, anthropomorphism, and uncanniness – Frontiers in Psychology
  4. Behavioral Assessment of a Humanoid Robot When Attracting Pedestrians in a Mall (International Journal of Social Robotics, 2022) – Europe PMC
  5. Novelty Knows No Boundaries: Why a Proper Investigation of Novelty Effects Within SHRI Should Begin by Addressing the Scientific Plurality of the Field – Frontiers in Robotics and AI
  6. Tour guide robot: a 5G-enabled robot museum guide – Frontiers in Robotics and AI
  7. Assessing pedestrian responses to autonomous and personal mobility robots in crowded public spaces (Science Advances, 2026) – Europe PMC
  8. Attitudes towards the impact of digitisation and automation on daily life – European Commission

博客更多内容

您的下一场活动想要一台机器人吗?

只需一次性说明日期、地点以及机器人需要完成的任务。供应商会给您回复报价,方便您逐项对比。

发送需求

您从事机器人行业吗?欢迎为我们的博客撰稿。 →