Robots on stage and in galleries: what museums, theatres and Expo 2025 actually showed

Humanoid robots in cultural venues mostly do one of three jobs: they talk to visitors as a permanent exhibit, they perform short scripted plays, or they serve as the centrepiece of a themed exhibition. Named examples include Ameca at museums in Nuremberg, Paderborn and Dubai, the robot theatre at Warsaw's Copernicus Science Centre, a May 2026 play in Istanbul and the android pavilion at Expo 2025 Osaka.

What these projects have in common is that the robots stay in one place, work to a script or a conversational AI system, and are surrounded by staff and a carefully prepared space. Below is what each project has published about itself, and what that tells event organisers.

Ameca as a museum exhibit that talks back

The robot that appears most often in museums is Ameca, made by the British company Engineered Arts. On its Ameca product page, the manufacturer lists 61 actuated degrees of freedom, cameras in the eyes, microphones and a chest speaker, and a cloud-based software suite called Tritium. The same page states plainly that Ameca cannot walk; it is a stationary robot built for expression and conversation.

Deutsches Museum Nürnberg

The Zukunftsmuseum (Future Museum) in Nuremberg, a branch of the Deutsches Museum, first showed Ameca as a guest exhibit in May 2022. According to the museum's page on Ameca in the Zukunftsmuseum, the robot became a permanent part of the collection on 9 July 2024. Visitors can hold conversations with it in several languages, from small talk to questions about the exhibition, and the museum says the conversation side is connected to ChatGPT-4o. The museum also mentions plans for robot shows and yoga sessions.

The same page notes that the robot's face now has 27 motors, up from 11 in the earlier version. That detail matters: most of what visitors respond to is the face, not the body.

Heinz Nixdorf MuseumsForum, Paderborn

The Heinz Nixdorf MuseumsForum (HNF), a computer museum in Paderborn, describes its own Ameca in a page titled Ameca – New at the HNF. The museum gives the robot's height as 1.87 metres and its weight as 62 kilograms, with 61 motors in total and 27 of them for facial expressions. It also points out that the legs are rigid, so the robot cannot move around the room, and that the face is grey and hairless so it is not assigned a human skin colour.

The HNF page also reports a figure from Engineered Arts: 29 Ameca units in use worldwide, mainly in museums and science centres, but also in research institutes and universities. The page does not date that figure, so treat it as a snapshot rather than a current count.

Museum of the Future, Dubai

In Dubai, Ameca has been greeting visitors on the Museum of the Future's 'Tomorrow, Today' floor since October 2022. Gulf News reported on 24 April 2025 that an upgraded version was unveiled during Dubai AI Week, speaking Arabic, English, Chinese, Hindi, Spanish and Russian. The report says the new version runs on Tritium 3 and adds facial and voice recognition, real-time translation and analysis of visitors' facial expressions and emotions.

That last feature is worth pausing on. Inside the EU, systems that infer emotions are restricted in some settings, which we explain in our article on emotion-reading robots and the AI Act. Any robot with cameras and microphones in a public space also raises data protection questions; see our guide to cameras, microphones and visitor privacy. This is general information, not legal advice.

Robots as actors: Warsaw's Robotic Theatre

The longest-running example of robots as performers is at the Copernicus Science Centre in Warsaw. Its Robotic Theatre page describes three humanoid actors of the RoboThespian type, built by Engineered Arts and moved by compressed air. The robots gesture, show emotion and speak several languages, with voices recorded by professional actors.

The current repertoire consists of two shows. One adapts a short story by Stanisław Lem, Prince Ferrix and Princess Crystal. The other, The Secret of an Empty Drawer, or The Ghosts of the Fourth Dimension, is based on Edwin A. Abbott's Flatland. Each lasts 20 minutes, entry is included in the general ticket, seats are first come, first served, and filming or photographing the performance is not allowed.

Copernicus also lists its other robots on a page about robots at its exhibitions. They include a RoboThespian from 2010 with serial number 0001, an Ameca from 2023 that acts as the 'curator' of an exhibition called The Future is Today, a robot made to resemble Nicolaus Copernicus, a Boston Dynamics Spot that visitors can steer, and a silicone robotic baby designed to provoke the 'uncanny valley' reaction. At the time we checked, the page marked Spot as unavailable for maintenance, a reminder that even permanent installations have downtime.

What the Warsaw shows illustrate is that robot theatre works best as fully scripted, pre-programmed performance. The robots do not improvise; the voices are human recordings and the movements are fixed in advance. That is a different product from a robot that chats with a crowd.

A humanoid in a cast of human actors: Istanbul, 2026

A newer format puts one robot into an otherwise human cast. The Turkish news site Diken reported on 18 May 2026 on the premiere of Şimdi Telefonunuza Bir Kod Gelecek (roughly, A Code Will Now Arrive on Your Phone), subtitled A Robotic Love Story, at the Atatürk Cultural Center (AKM) in Istanbul. The play was written and directed by Metin Yıldız, who also acts in it alongside two other human performers.

In the story, a divorced man receives a package containing a humanoid copy of his former wife, played by the robot. The report credits the company LTC İnovasyon with the robot's technical set-up but does not name the robot model or say how much of its performance is autonomous, pre-programmed or controlled by an operator.

The production described itself as the first theatre play in the world in which an AI-powered humanoid robot takes an actor's role. That claim depends on how you define 'AI-powered' and 'actor': robots have been performing scripted plays in Warsaw for years. We could not verify the claim and report it as the production's own description.

Androids as exhibition: Expo 2025 Osaka

At world-expo scale, robotics researcher Hiroshi Ishiguro of Osaka University produced the 'Future of Life' signature pavilion at Expo 2025 Osaka, Kansai. In a preview released by the Expo organisers, the pavilion was described in three zones: the history of Japanese life in objects, everyday life 50 years from now, and human evolution 1,000 years ahead.

The organisers said around 50 avatars were created for the pavilion, about 30 of them on display, and introduced named robots and androids including Petra, Punica, Pangie, Yamatoroid and the child-like Asukaroid. Here the robots were not performers or guides but the exhibition itself: visitors walked through staged scenes in which androids represented a future society.

What the audience actually saw

Across these projects, a few patterns are consistent:

• Faces over feet. Ameca, RoboThespian and the Expo androids are valued for expression and speech. None of the Ameca installations above involve walking, and the HNF says so directly.
• Script or conversation, rarely both. Theatre uses fixed programmes and recorded voices; museum exhibits use large language models for open conversation. Our guide to teleoperated and autonomous robots explains how audiences perceive the difference.
• Fixed position, controlled space. Robots sit on stages or plinths with seating, barriers or staff nearby, rather than moving freely through crowds.
• Language as a selling point. Nuremberg, Paderborn and Dubai all stress multilingual conversation, which suits international audiences but depends on microphones coping with noisy rooms.

What this means for event organisers

• Decide whether you want a performer or a conversation partner. A scripted show needs content created in advance; see our guide on custom programming and choreography lead times. A conversational robot needs a quiet enough spot and a clear topic.
• Plan for sound. Museums can control acoustics; a busy foyer cannot. Read our notes on speech, languages and noisy venues before promising multilingual chat.
• Expect static installations for expressive humanoids. If you need a robot that walks among guests, that is usually a different machine from the one with the most expressive face.
• Check what the robot records. If it uses cameras, face or emotion analysis, clarify data handling before the event.
• Brief suppliers with the format, not just the robot. Describe the audience, duration, script and space when you send one request to several suppliers.

Frequently asked questions

Which humanoid robot is most common in museums?

Among the examples we checked, Ameca by Engineered Arts appears most often. It is a permanent exhibit at the Deutsches Museum Nürnberg and the Heinz Nixdorf MuseumsForum in Paderborn, at the Museum of the Future in Dubai, and at the Copernicus Science Centre in Warsaw. The HNF cites a company figure of 29 Ameca units in use, mainly in museums and science centres, though that figure is not dated.

Do robot actors improvise on stage?

In the long-running Warsaw example, no. The Copernicus Science Centre's RoboThespians perform fixed 20-minute shows with movements programmed in advance and voices recorded by human actors. The 2026 Istanbul play describes its robot as AI-powered, but the published report does not explain how much of the performance is autonomous, pre-programmed or operator-controlled, so that remains unclear.

Can Ameca walk around an exhibition or event?

No. Engineered Arts states on its product page that Ameca cannot walk, and the Heinz Nixdorf MuseumsForum notes that the robot's legs are rigid. Ameca is designed for facial expression, gesture and conversation from a fixed position. Organisers who need a robot that moves through a venue usually have to look at different machines, which tend to be less expressive.

Are there privacy issues with talking robots in museums?

Potentially. Conversational robots use microphones and often cameras, and some versions add face or voice recognition; the upgraded Ameca in Dubai is reported to analyse facial expressions and emotions. In the EU, data protection rules and the AI Act's limits on emotion recognition may apply depending on the setting. Organisers should ask what is recorded, processed and stored. This is general information, not legal advice.

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 2026-11-02. Spotted an error? Write to info@humanoid4rent.com.