Kanban StudiosKanban Studios
SELECTED WORK · 03

Guide Bot - the guide.

An autonomous indoor guide robot with a voice-driven 3D-avatar interface, built for an advanced-technology learning, innovation, and demonstration centre in Abu Dhabi. It patrols mapped routes and self-charges on its own, while visitors ask for directions, facility info, and programme details and the bot replies by voice through a lip-synced 3D avatar.

Visit the live appbridgebot.vercel.app

Built for EDGE Group P.J.S.C

THE APP

BRIDGE AI Bot - the kiosk interface.

Tap SPEAK, ask a question, and the lip-synced 3D avatar answers out loud - the same screen visitors meet at the centre.

Guide Bot - the guide robot on the floor of the lab
Guide Bot app - light mode interfaceGuide Bot app - dark mode interface
WHAT IT DOES

Ask it anything about the centre.

Directions, facilities, and programmes - Lean, Six Sigma, Lean Digital, Agile, Data Analytics, IoT, CEO 4.0, the Engineering Bootcamp, and more - answered out loud by a 3D avatar, and the robot can lead visitors there on mapped routes.

01

Autonomous navigation

ROS-based navigation (NAV2) lets the robot map the space, patrol set routes, and lead visitors from A to B on its own - no rails, no operator driving it.

02

Self-docking charge

When the battery runs low the robot returns to its dock and recharges itself, so it stays on the floor and ready without anyone plugging it in.

03

Lip-synced 3D avatar

A muhammad.glb avatar rendered with Three.js speaks each reply with synced lip movement, so visitors talk to a face, not a text box.

04

Voice in, voice out

SpeechRecognition takes the visitor's question and text-to-speech reads the answer aloud - fully hands-free at the kiosk.

05

Instant local FAQ lookup

A local faqs.json knowledge base answers directions (VIP room, offices, classrooms) and facility questions instantly, with no round-trip.

06

LLM fallback

Anything not matched locally falls through to OpenAI (gpt-4o-mini), proxied through the backend so the API key stays server-side.

HOW IT FLOWS
Voice inVisitor asks
PerceptionSpeech to text
FAQ lookupLocal faqs.json
LLM fallbackgpt-4o-mini
Voice outText to speech
ScreenAvatar lip-sync
BUILT WITH
ROSNAV2PYTHONTHREE.JSOPENAISPEECH RECOGNITIONTEXT-TO-SPEECHEXPRESSVERCEL
Chat on WhatsAppWhatsApp