An introduction to Keeper

Six small chapters. A whole world of stories.

Chapter 1 of 6

More than a label.

Tap play to begin with sound.

0:001:28

Lisieux TitanVaikom, KeralaWRO 2026 · Future Innovators

Keeper of Stories

A robot that maps a museum, discovers the exhibits inside it and brings their stories to life.

Qualified for APAC · Ranked 16th
Sees
Beyond art styles
Maps
With LiDAR
APAC
Ranked 16th
Meet the APAC edition The APAC Keeper, with bright blue animated eyes, a camera above its face, a tablet, soft arms and musical notes across its cream body. The team and Keeper at the APAC booth.Our little corner A group photo with people from Korea at APAC.A world of stories
Keeper of Stories · Upgraded for APAC 2026

01Why we built it

Most paintings in a museum get a small card: a name, a place, a date. The story behind a Warli dance or a Madhubani peacock doesn’t fit on it.

Folk painting traditions like Warli, Madhubani and Pattachitra are still practised by families and communities across India. Every shape in them means something. But when you see one on a wall, you usually get a title and not much else.

The theme of the World Robot Olympiad 2026 was Robots Meet Culture. We asked what a robot could actually do for culture, and landed on something simple: stand beside a painting, explain it clearly, and answer whatever people want to know.

“Because heritage deserves more than a label.”

From our exhibition poster at Nationals

02Try it

Pick a painting. Ask Keeper about it.

These five paintings come from Keeper’s original repertoire of 15+ folk art styles. Choose one for a glimpse of how a conversation starts. The upgraded Keeper now explores far beyond that original collection.

Recognised Madhubani · Mithila, Bihar

You

What am I looking at?

Keeper

This is a Madhubani painting, from the Mithila region of Bihar. It shows a peacock with a fish curled into its body, two favourite Madhubani subjects. Notice the double outlines, and how almost no space is left empty.

Ask a follow-up

A scripted preview of the conversations Keeper has at exhibitions. The real robot hears spoken questions, works out its own answers and speaks them aloud. Answers here are short summaries; the artists and communities behind each tradition are the real experts.

03How it works

It looks. It listens. It answers.

A camera, a voice and a map of the space around it. Together, they let Keeper turn what it finds into a conversation.

  1. 01

    Sees

    The camera finds paintings, sculptures, objects and signs. Visual AI describes what it sees, while text recognition reads the museum’s own labels. It can learn about unfamiliar exhibits without being trained on a fixed list of art styles.

  2. 02

    Explains

    Keeper combines what it sees with the exhibit’s label and its saved museum knowledge. It shows material on its screen and explains what to look for. Uncertain details go to a curator for review.

  3. 03

    Listens

    Ask a question out loud. Speech recognition turns it into text, a language model writes the answer, and Keeper says it back.

  4. 04

    Knows the way

    LiDAR builds a map of walls and open space. Keeper remembers where exhibits are and plans a route when a visitor asks to see one. Mapping and movement run locally; conversation also has a local AI fallback.

And it comes to you.

Keeper doesn’t wait for people to walk up. It drives over to visitors on its own and starts the conversation by asking if there’s anything they’d like to know.

Keeper, rolling up“Hello! Is there anything you’d like to know?”

Who moves Keeper

AIWheels and arm
The AI can request a destination or an expressive gesture. Keeper’s robotics software turns that request into movement, using the map and sensors.
CodeHome-made routines
Not every move comes from the AI. Some are still run by code we wrote, like the routines behind Keeper’s other home-made autonomous features.

What’s inside

The system behind the upgraded Keeper.

Computing
Laptop for AI and museum knowledge; Raspberry Pi for the robot’s camera, microphone and navigation
Motors & sensors
Arduino
Code
Python, C, C++ and Bash
Vision & mapping
LiDAR SLAM, visual AI and OCR for exhibit labels
Language models
Sarvam, with local Qwen as a fallback
Voice
Speech recognition and text-to-speech, with local speech support
Museum memory
Exhibit records linked to their location, sources and curator corrections
Body
An animated face, a tablet, expressive arms and a wheeled base

04The build

It started as a LEGO car with wires everywhere.

The first Keeper fit on a tabletop, and it’s the one we took to Regionals in Kochi. It already worked: it ran the same AI, answered spoken questions and moved when the AI told it to. It placed 3rd. After that we built the full-size Keeper for Nationals.

Top view of the LEGO prototype we took to Regionals: four yellow motors and wheels on a LEGO frame, a blue Arduino board, a small ESP32-CAM and jumper wires everywhere. Front view of the same prototype: a LEGO frame on a small four-wheeled base with yellow motors and exposed red, black and green wires.
v0.1The Regionals prototype. A LEGO frame, four motors, a small ESP32-CAM and a lot of jumper wires. It had no arms and couldn’t go up to visitors on its own.
The full-size Keeper we built for Nationals, standing at human height with its screen, LED eyes and padded arms.
v1.0The full-size Keeper. Built after Regionals for Nationals. Human height, with arms, a proper webcam, a screen and LED eyes. It understands its surroundings better and drives up to visitors on its own.
  1. Vaikom

    The idea

    Could a robot help people understand the art in front of them?

  2. Vaikom

    LEGO prototype

    A small LEGO robot that already answered spoken questions and moved when the AI said.

  3. Kochi

    Regionals

    We took the LEGO prototype and placed 3rd, earning a place at Nationals.

  4. Vaikom

    Full-size Keeper

    We rebuilt it at human height, with arms, a better camera and the smarts to drive up to visitors on its own.

  5. Hyderabad

    Nationals

    26–28 August 2026 at GMR Arena.

  6. APAC

    Ranked 16th

    7th at Nationals, then qualification for the Asia-Pacific competition.

05WRO India National Championship

7th in India.

After our LEGO prototype placed 3rd at Regionals, we built the full-size Keeper and took it to GMR Arena in Hyderabad for the WRO India National Championship in August 2026. The theme was Robots Meet Culture, the exact question Keeper was built to answer. We finished seventh in the Future Innovators Senior category.

Regionals · Kochi
3rd
Nationals · Hyderabad
7th
Dates
26–28 Aug
The WRO welcome board: National Championship 2026, Robots Meet Culture, 26–28 August, GMR Arena, Hyderabad.
The welcome board at GMR Arena

06The APAC chapter

Vaikom, Kerala ↗ Asia-Pacific

Small town.
Bigger possibilities.

Built in Vaikom. Rebuilt for APAC. Our museum guide came back with a bigger mind, a bigger stage and a world of new connections.

Qualified · Asia-Pacific

16th

Ranked in APAC
Team 4076INDIA ↗
Keeper of Stories / APAC build The upgraded Keeper: blue animated eyes, camera, visitor tablet, padded arms and a cream body decorated with musical notes.

Touch a note. Meet the new Keeper.

The upgrade / beyond 15 art styles

A painting was the start.
Now, a whole museum.

LiDAR + autonomous exploration

First, learn the space.

Keeper explores by itself, mapping walls and open floor with LiDAR. It chooses where to look next and keeps track of its position as the museum takes shape.

From a floor plan to a museum Keeper understands An illustrative museum with three galleries. LiDAR maps walls and open space; camera observations add exhibits and labels; a route connects Keeper to an exhibit. GALLERY 01GALLERY 02GALLERY 03 COLLECTIONSCULPTUREENTRANCE KEEPER
01 / Building a floor plan · Process illustration

From APAC

Lisieux Titan at the APAC booth.

07At the booth

Three days of explaining Keeper to anyone who stopped.

Judges, other teams, parents, curious kids. Our posters covered what Keeper does, the tech inside it and where we want to take it next. The best part was watching people start asking it questions.

Lisieux Titan students at their booth, pointing to posters about Keeper of Stories, its tech stack and the art traditions it supports.
Our booth at WRO India 2026

08The team

Lisieux Titan

Three students from Lisieux English School in Vaikom, Kerala, plus the coach and mentor who kept them going. Keeper took months of building, breaking and rebuilding.

Lisieux Titan outside GMR Arena, Hyderabad. Francis Joseph takes the selfie in front; behind him are Baiju R (Goutham B's father) in a cap, Jini Joseph (Francis's mother), Goutham B, Gautham R, mentor Divya M U and supporter Ajeesh A in a green shirt.
Outside GMR Arena, Hyderabad, with family and supporters. Tap a name to find them.

The students

  1. 1Software & AI lead

    Francis Joseph

    Led the code that lets Keeper recognise paintings, answer questions and speak.

  2. 2Design

    Goutham B

    Shaped how Keeper looks, and shared hardware and software work with Gautham R.

  3. 3Hardware

    Gautham R

    Built the physical robot, and shared design and software work with Goutham B.

Behind the team

  • –Coach

    Reshma P R

    Coached the team through the build. Not in this photo.

  • 4Mentor & coordinator

    Divya M U

    Mentored the team and travelled with them to Hyderabad.

Also in the photo: Jini Joseph (Francis’s mother), Baiju R (Goutham B’s father) and Ajeesh A, supporters who backed the team all the way.

Lisieux English School, Vaikom crest

Lisieux English SchoolVaikom, Kerala, India

09What’s next

Where Keeper could go from here.

The APAC upgrades bring Keeper closer to the places it was built for. Our next steps should grow together with the artists, curators and communities whose stories it shares.

  1. A

    More languages

    So people can ask questions in the language they think in, not just English.

  2. B

    Deeper stories

    Pair what Keeper discovers with the knowledge of artists and curators, so every explanation carries more context.

  3. C

    Guided art-making

    Helping visitors try a technique themselves, not just hear about it.

  4. D

    Real museums and schools

    Anywhere a painting deserves more than a label.

From art to heart.

The line on our poster, and the whole idea in four words.