We give India’s heritage
an autonomous guide.

Yantrax is an AI-powered robot that guides museum visitors, maps archaeological sites, and digitizes historical manuscripts — speaking multiple Indian languages, completely offline, entirely on its own.

Meet Yantrax

Built for the World Robot Olympiad

The Robot

One Robot, Every Role
a Museum Needs

India holds thousands of years of history — and far too few hands to guide, map and preserve it all. Yantrax is our answer.

Inside a museum, Yantrax drives itself between exhibits, greets visitors it detects along the way, and explains each artifact through speech and an interactive touchscreen — in the visitor’s own Indian language. At archaeological sites, the same robot switches roles: it maps the terrain, records what it sees, and turns fragile handwritten manuscripts into searchable digital text with its onboard OCR.

No internet. No operator. One intelligent platform, preserving culture while it guides the people who came to experience it.

See the tech
The Yantrax robot — a cream-colored autonomous museum guide decorated with traditional Indian folk art, with a stereo camera head, marigold garland, and a touchscreen greeting visitors with Namaste
Yantrax // Prototype 01

System Architecture

Perceive. Understand. Act.

Every ability runs on one Jetson Orin Nano 8GB under ROS 2 Humble — no cloud service anywhere in the loop.

Perceive

  • YOLO26n · TensorRTspots people at 53 FPS, INT8
  • ByteTracklocks onto your track ID, not the nearest person
  • RPLidar A1360° laser scan for mapping & planning
  • HC-SR04forward cutoff, enforced in firmware
  • 2× USB camerasperson detection + document chamber

Understand

  • Wake word“Yantrax” — or a double clap
  • whisper.cppspeech to text, on-device
  • Qwen2.5-3Boffline LLM, answers in 9 languages
  • Pipernatural text-to-speech
  • Tesseractmanuscript & document OCR

Act

  • Follow-the-gapreplans a heading through the best gap in the scan
  • SLAM + live mapoccupancy grid streams to the dashboard
  • Arduino Megadifferential drive base
  • E-stop + watchdoglatching stop, 400 ms firmware watchdog

Mission & Vision

Preserving Heritage Through
Intelligent Robotics

Get in touch

Every year, manuscripts fade, inscriptions weather away, and stories go untold because there aren’t enough guides, archivists and surveyors to go around. Our mission is to put intelligent robotics to work where human hands are scarce — guiding visitors, mapping sites and digitizing documents before time takes them.

We imagine museums where every visitor is welcomed in their own language, and archaeological sites that are mapped and archived as carefully as they were built. Yantrax is our first step toward that vision: technology that doesn’t replace the human connection to heritage, but multiplies it.

What Yantrax Does

Offline Intelligence

Every Ability, Onboard.
No Internet Required.


Everything Yantrax knows runs on a single NVIDIA Jetson Orin Nano. Autonomous navigation steers it through crowded halls; computer vision spots visitors and artifacts; speech recognition understands questions in multiple Indian languages and answers in kind. Because none of it needs the cloud, Yantrax works just as well in a basement archive as in a modern gallery.

The same platform carries an OCR pipeline that photographs historical manuscripts and converts them into digital, searchable text — and a mapping stack that surveys archaeological sites as it moves. Touch, voice, vision and motion, fused into one machine that any museum can simply switch on.

Get in touch

Measured, Not Estimated

Numbers From the Bench,
Not the Brochure

0 FPS person detection, steady state
0% OCR character accuracy
0 s mean document scan time
0 languages, fully offline
0 ms firmware watchdog
0 build warnings, cloud calls, excuses

Say This

Talk to It Like
a Person

Yantrax never speaks first — person detection only decides when it is allowed to listen. Wake it with its name or a double clap, then just talk.

“Yantrax”wakes it — so does a double clap

“Follow me”locks onto you, routes around obstacles

“Map this area”builds and saves the SLAM map

“Read this document”scans it, then reads it aloud

“Status report”spoken health check of every subsystem

“Stop”latching e-stop — it stays stopped until released

Anything elseanswered by the offline LLM, in your language

Hardware

The Full Bill of Brains

PartWhat it does
Jetson Orin Nano 8GBruns every model onboard — roughly 4.5 GB usable in practice
Arduino Mega 2560drive base, latching e-stop, 400 ms watchdog
2× BTS7960 + 12V gearmotorsdifferential drive
RPLidar A1laser scanning for SLAM and planning
USB cameraperson detection
USB camera + LED chamberdocument scanning rig
HC-SR04 ultrasonicforward obstacle cutoff, enforced in firmware
23″ portrait displaythe dashboard is the whole UI
View the code on GitHub

Every ROS 2 package builds with zero warnings. The preflight script verifies every subsystem, in order.

Milestones

From Idea to the
Olympiad Stage

Phase 01 // The Spark

One guide, eighty visitors

It started with a school museum visit and a simple observation: one guide, eighty visitors, four languages in the room. We formed Team Yantrax around a question — what if the museum itself could speak to everyone?

  • Team formed
  • Problem defined

Phase 02 // The Build

A museum brain on one board

Months of building followed: an autonomous drive base, a touchscreen interface, multilingual speech, person detection, site mapping and a manuscript OCR pipeline — all squeezed onto one Jetson Orin Nano and proven working completely offline.

  • Jetson online
  • Nav stack
  • OCR v1
  • 9 languages

Phase 03 // The Stage

Destination: WRO Future Innovators

Now Yantrax carries our work to the World Robot Olympiad Future Innovators, Senior Category — representing our belief that the best way to protect India’s past is to build boldly for its future.

  • Senior category
  • Team of three
  • One robot

The Robot, Every Angle

Designed Like an Artifact,
Built Like a Machine

Traditional Indian folk art on the outside, a full autonomy stack on the inside — because a robot that lives among heritage should look like it belongs there.

Team Yantrax

The People Behind the
Robot

Team Yantrax is three students who believe heritage and high technology belong in the same sentence. Between us we cover the robot end to end — the autonomy that moves it, the AI that lets it see and speak, and the design that makes a machine feel welcome in a museum.

We compete in the World Robot Olympiad Future Innovators, Senior Category, under one motto: “Preserving Heritage Through Intelligent Robotics.”

Daivik Reddy

Team Member

“A robot in a museum shouldn’t feel like an exhibit. It should feel like the friendliest guide you’ve ever had.”

DR

Jeshwanth

Team Member

“Offline is the hard mode of AI — and exactly why it works in every museum, not just the connected ones.”

J

Juhitha

Team Member

“Every manuscript we digitize is a story that can never be lost again. That’s worth every late night.”

J

Get in touch

Let’s Preserve the Past,
Together

Maybe you run a museum or heritage site that could use an untiring, multilingual guide. Maybe you’re a mentor, judge or fellow builder curious about how Yantrax works under the hood. Or maybe you simply love the idea of robots keeping culture alive. Whatever brings you here, write to us — we’d love to talk.