Skip to content

Repository files navigation

🥾 WalkInMyShoes — Immersive Disability Empathy & Accessibility Training Platform

Built for AWS AI for Bharat 2026 Hackathon "Step into someone else's world. Change yours forever."

Live Demo GitHub AWS


🇮🇳 Why This Matters for Bharat

India has 2.68 crore people living with disabilities (Census 2011) — yet accessibility remains an afterthought in most public and private spaces. The Rights of Persons with Disabilities (RPWD) Act 2016 mandates inclusive infrastructure, but enforcement is minimal due to lack of awareness, training, and affordable auditing tools.

WalkInMyShoes bridges this gap by combining immersive VR empathy simulations with AI-powered real-time accessibility auditing — making empathy training experiential and compliance auditing accessible to any building owner with a smartphone.


🎯 The Problem

Challenge Scale
People with disabilities in India 2.68 crore (Census 2011)
Buildings compliant with RPWD Act 2016 <5%
Organizations with accessibility training <10%
Traditional training effectiveness 20% retention (lecture-based)
Cost of professional ADA/RPWD audit ₹50,000–₹2,00,000

Traditional accessibility training relies on passive videos and lectures. Organizations struggle with compliance because employees have never experienced what inaccessibility actually feels like. Professional accessibility audits are expensive and inaccessible to small businesses.


💡 Our Solution

An immersive, AI-powered WebXR platform running entirely on AWS that:

  1. VR Simulations — Let users experience the world through the eyes of people with disabilities
  2. AR Accessibility Auditor — Scan any real-world space with a smartphone camera for instant ADA/RPWD compliance analysis
  3. AI Expert Guide — Powered by Amazon Bedrock (Nova 2 Lite) for real-time accessibility guidance
  4. Impact Dashboard — Measure empathy growth, earn certificates, track organizational compliance

Works on any device — no app installation, no VR headset required.


🏗️ AWS Architecture (7 Services)

┌─────────────────────────────────────────────────────────────────────┐
│                          USER (Any Device)                          │
│             Browser | Mobile | Tablet | Desktop                     │
└──────────────────────────────┬──────────────────────────────────────┘
                               │ HTTPS
                               ▼
┌─────────────────────────────────────────────────────────────────────┐
│                    1. AMAZON CLOUDFRONT (CDN)                       │
│  Distribution: E2FKI267871EMW | Region: Global Edge                 │
│  • HTTPS enforcement  • Asset caching  • Gzip compression           │
│  • Domain: d2d1ibzdtgm1nq.cloudfront.net                           │
└──────────────────┬───────────────────────┬──────────────────────────┘
                   │                       │
                   ▼                       ▼
┌─────────────────────────┐   ┌────────────────────────────────────────┐
│  2. AMAZON S3           │   │  3. AMAZON API GATEWAY                 │
│  walkinmyshoes-         │   │  ID: x5rbnqm2v4                        │
│  frontend-2026          │   │  Stage: /dev                           │
│  • React app bundle     │   │  Routes:                               │
│  • Static assets        │   │    POST /ai   → Lambda (AI)            │
│  • index.html (no-cache)│   │    GET  /analytics                     │
└─────────────────────────┘   │    POST /progress                      │
                               │    GET  /leaderboard                   │
                               └──────────────┬─────────────────────────┘
                                              │
                                              ▼
                               ┌─────────────────────────────────────────┐
                               │  4. AWS LAMBDA                          │
                               │  Function: walkinmyshoes-ai             │
                               │  Runtime: Node.js 20.x                  │
                               │  Memory: 256MB | Timeout: 60s           │
                               │  Handler: index.handler                 │
                               │                                         │
                               │  Actions:                               │
                               │  • analyzeAccessibility (AR scan)       │
                               │  • editImage (Synthesize Fix)           │
                               │  • chat (AI Guide conversation)         │
                               └────────┬────────────────┬───────────────┘
                                        │                │
                    ┌───────────────────┘                └──────────────────┐
                    ▼                                                        ▼
     ┌──────────────────────────┐                        ┌──────────────────────────────┐
     │  5. AMAZON REKOGNITION   │                        │  6. AMAZON BEDROCK           │
     │  • DetectLabels          │                        │  Model: Nova 2 Lite          │
     │    (40 labels, 55% conf) │                        │  ID: us.amazon.nova-lite-v1:0│
     │  • DetectText            │                        │  • ADA Accessibility Audit   │
     │    (bounding boxes)      │                        │  • Remediation Reports       │
     │  • 4MB image max         │                        │  • Expert Chat Responses     │
     │  • Parallel processing   │                        │  • RPWD Act 2016 context     │
     └──────────────────────────┘                        └──────────────────────────────┘
                                        │
                                        ▼
                        ┌───────────────────────────────┐
                        │  7. AMAZON COGNITO            │
                        │  Pool: us-east-1_VHC0a9xM3   │
                        │  • User registration/login    │
                        │  • JWT token management       │
                        │  • Session handling           │
                        │  • Hosted UI for auth flow    │
                        └───────────────────────────────┘

AWS Services Summary

# Service Role Resource
1 Amazon CloudFront Global CDN, HTTPS, caching E2FKI267871EMW
2 Amazon S3 Static frontend hosting walkinmyshoes-frontend-2026
3 Amazon API Gateway REST API routing x5rbnqm2v4 (us-east-1)
4 AWS Lambda Serverless AI orchestration walkinmyshoes-ai
5 Amazon Rekognition Computer vision on camera frames DetectLabels + DetectText
6 Amazon Bedrock LLM for audit/chat/reports Nova 2 Lite (us.amazon.nova-lite-v1:0)
7 Amazon Cognito User authentication & sessions us-east-1_VHC0a9xM3

🤖 AI Pipeline: How the AR Auditor Works

Camera Frame (JPEG/PNG)
        │
        ▼
  Base64 Encode
        │
        ▼
  API Gateway /ai  ──→  Lambda (walkinmyshoes-ai)
                                │
                    ┌───────────┴───────────┐
                    │                       │
                    ▼                       ▼
            Rekognition                Rekognition
            DetectLabels               DetectText
            (objects+confidence        (visible text,
            +bounding boxes)           signs, numbers)
                    │                       │
                    └───────────┬───────────┘
                                │
                                ▼
                    Scene Description Assembly
                    (labels + text + bounding
                     box coordinates 0-1000)
                                │
                                ▼
                    Amazon Bedrock Nova 2 Lite
                    (ADA 2010 + WCAG 2.1 +
                     RPWD Act 2016 audit prompt)
                                │
                                ▼
                    JSON Audit Result
                    {issues[], overallComplianceScore}
                                │
                                ▼
                    AR Overlay on Camera Feed
                    (color-coded bounding boxes,
                     compliance indicators, cost estimates)

🎮 Features

1. VR Disability Simulations

Simulation What You Experience Key Learning
Visual Impairment 5 stages of vision loss — blur to blindness 83% of websites lack sufficient color contrast
Hearing Loss Frequency loss, tinnitus, no captions 466M people worldwide have disabling hearing loss
Motor Disability Wheelchair physics, architectural barriers Users take 3–5× longer to complete tasks
Color Blindness Protanopia, Deuteranopia, Tritanopia, Achromatopsia 1 in 12 men are colorblind

Each simulation includes:

  • Immersive 3D environment rendered in browser (Three.js + WebGL)
  • AI Guide panel for real-time Q&A
  • Contextual tasks that reveal barriers
  • Debrief with real-world statistics
  • Empathy score measurement

2. AR Accessibility Auditor

Point your phone camera at any space for instant compliance analysis:

  • Rekognition detects objects, doors, ramps, signs, pathways with confidence scores and bounding boxes
  • Nova 2 Lite via Bedrock performs a full ADA 2010 + WCAG 2.1 + RPWD Act 2016 audit
  • Color-coded overlays show COMPLIANT (green), NON_COMPLIANT (red), WARNING (yellow)
  • Each issue includes description, ADA code reference, cost estimate (₹ and $), and exact coordinates
  • Minimum 5 issues detected per scan with overall compliance score (0–100)

Audit Categories:

  • Doorway width (ADA: ≥32", ideal 36")
  • Ramp slope (ADA: ≤1:12 ratio)
  • Pathway width (ADA: ≥36")
  • Tactile paving (truncated domes at hazard transitions)
  • Signage contrast (≥3:1 ratio, Braille, 60" AFF centerline)
  • Controls reach range (15–48" AFF, lever hardware)
  • Flooring (ASTM ≥0.6 wet slip resistance)
  • Lighting (≥50 foot-candles at task areas)
  • RPWD Act 2016 Indian-specific requirements

3. Synthesize Fix (AI Remediation)

  • Click any detected issue → "Synthesize Fix" generates a complete contractor-ready remediation report
  • Powered by Amazon Bedrock Nova 2 Lite + Rekognition scene context
  • Report includes: materials, installation steps, cost breakdown in ₹ and $, ADA code sections, priority rating, timeline
  • Sections: Site Observation → ADA Violation → Remediation Spec → Expected Outcome → Cost → Code Refs → Timeline

4. AI Expert Guide

  • Context-aware conversational assistant powered by Amazon Bedrock
  • Integrated into every simulation and the AR Auditor
  • Knows which scenario you're in, what issues were detected
  • Cites exact ADA sections (§) and WCAG criteria (SC X.X.X)
  • References RPWD Act 2016 and NBC 2016 for Indian context
  • Uses India-specific statistics
  • Multi-turn conversation with full history maintained per session

5. Impact Dashboard & Certification

  • Empathy score algorithm (0–100) based on behavioral data
  • Pre/post assessment quizzes tracking knowledge gain
  • Downloadable PDF certificates with completion date and score
  • Leaderboard rankings
  • Organization-wide analytics (future)
  • Real-world impact statistics

🛠️ Technical Stack

Frontend

React 18 + TypeScript
Vite 5 (build tool)
Three.js (3D/WebXR rendering)
Tailwind CSS (styling)
WebXR Device API (VR/AR)
Web Audio API (hearing simulation)
SVG Filters (color blindness simulation)

Backend (AWS Lambda — Node.js 20.x)

@aws-sdk/client-bedrock-runtime
@aws-sdk/client-rekognition
Amazon Bedrock: us.amazon.nova-lite-v1:0
Amazon Rekognition: DetectLabels + DetectText
CORS: wildcard (*) for CloudFront + localhost

Infrastructure

Region: us-east-1
CloudFront Distribution: E2FKI267871EMW
S3 Bucket: walkinmyshoes-frontend-2026
API Gateway: x5rbnqm2v4
Lambda: walkinmyshoes-ai (256MB, 60s timeout)
Cognito: us-east-1_VHC0a9xM3

📁 Project Structure

wims_bedrock_final/
├── components/
│   ├── AIGuide.tsx              # Bedrock-powered conversational assistant
│   ├── ARAuditor.tsx            # AR camera + Rekognition + Bedrock audit
│   ├── ColorBlindnessScene.tsx  # Color vision deficiency simulation
│   ├── HearingLossScene.tsx     # Hearing impairment simulation
│   ├── ImpactDashboard.tsx      # Empathy scoring & certification
│   ├── Layout.tsx               # App shell & navigation
│   ├── MotorDisabilityScene.tsx # Wheelchair navigation simulation
│   ├── Onboarding.tsx           # Tutorial flow
│   └── VisualImpairmentScene.tsx # Vision loss simulation
├── services/
│   ├── bedrock.ts               # Amazon Bedrock service (Nova 2 Lite)
│   ├── api.ts                   # Backend API client
│   └── auth.ts                  # Cognito authentication
├── lambda/
│   └── index.mjs                # Lambda handler (Rekognition + Bedrock)
├── App.tsx                      # Main application router
├── constants.tsx                # App-wide constants
├── types.ts                     # TypeScript type definitions
├── .env                         # Environment variables (pre-configured)
├── vite.config.ts               # Vite build configuration
├── package.json                 # Dependencies
└── index.html                   # App entry point

🚀 Getting Started

Prerequisites

  • Node.js 18+
  • AWS CLI configured with your credentials
  • AWS account with Bedrock + Rekognition access

Local Development

# 1. Install dependencies
npm install

# 2. Start dev server
npm run dev

# 3. Open browser
# http://localhost:5173

Environment Variables (pre-configured in .env)

VITE_API_BASE_URL=https://x5rbnqm2v4.execute-api.us-east-1.amazonaws.com/dev
VITE_COGNITO_DOMAIN=walkinmyshoes-auth.auth.us-east-1.amazoncognito.com
VITE_COGNITO_CLIENT_ID=20iorjt3l68ebjgohdrd0agj8a
VITE_COGNITO_USER_POOL_ID=us-east-1_VHC0a9xM3
VITE_COGNITO_REGION=us-east-1
S3_BUCKET=walkinmyshoes-frontend-2026
CLOUDFRONT_ID=E2FKI267871EMW
AWS_REGION=us-east-1

Production Deploy

# Build
npm run build

# Deploy to S3
aws s3 sync dist/ s3://walkinmyshoes-frontend-2026 --delete --region us-east-1 --cache-control "public,max-age=31536000,immutable" --exclude "index.html"

# Upload index.html (no-cache)
aws s3 cp dist/index.html s3://walkinmyshoes-frontend-2026/index.html --region us-east-1 --cache-control "no-cache,no-store,must-revalidate"

# Invalidate CloudFront
aws cloudfront create-invalidation --distribution-id E2FKI267871EMW --paths "/*"

Lambda Deployment

After any changes to lambda/index.mjs:

  1. AWS Console → Lambda → walkinmyshoes-ai
  2. Code tab → paste updated lambda/index.mjs
  3. Click Deploy

📊 Impact & Metrics

Social Impact

  • Addresses RPWD Act 2016 compliance gap in India
  • Makes accessibility auditing affordable (₹0 vs ₹50,000+ for professionals)
  • Trains empathy in 15–25 minutes vs 2-day workshops
  • 90% information retention vs 20% for passive lectures

Technical Performance

  • AR scan analysis: ~3–8 seconds per frame
  • AI Guide response: ~2–4 seconds
  • CloudFront global delivery: <200ms latency at edge
  • Lambda cold start: <2 seconds

Business Model

  • B2B SaaS: ₹4,000/user/year
  • AR Audit Reports: ₹2,500 average discovered compliance value per scan
  • Target: Corporates, hospitals, government buildings, educational institutions

🏆 Why WalkInMyShoes Wins

✅ Real AWS AI (Not Just Wrapper)

  • Amazon Rekognition does actual computer vision on camera frames — not just API calls
  • Amazon Bedrock Nova 2 Lite runs a genuine ADA/RPWD compliance audit from scene data
  • Two AI services working in parallel (Rekognition + Bedrock) per request

✅ India-First Problem

  • RPWD Act 2016 compliance is a real, massive problem in India
  • 2.68 crore affected Indians — larger than many countries' total populations
  • Cost-effective solution for a market that can't afford traditional audits

✅ Live, Working Demo

  • Judges can try it right now on any device
  • AR Auditor works on any smartphone with a camera
  • No login required to experience simulations

✅ Technical Depth

  • 7 AWS services properly integrated, not just listed
  • Serverless architecture that scales to millions of users
  • Real computer vision pipeline: camera → Rekognition → LLM → AR overlay

✅ Emotional Impact

  • Judges experience disability firsthand — not just read about it
  • Emotional memory drives lasting impression
  • Unique "wow factor" that no other hackathon project has

🗺️ Roadmap

Phase Timeline Features
Phase 1 March 2026 VR simulations, AR Auditor, AI Guide, Bedrock, CloudFront
Phase 2 Q2 2026 Mobile app, multi-language, enterprise dashboard
Phase 3 Q3 2026 Custom scenario builder, API integrations, white-label
Phase 4 Q4 2026 VR headset optimization, haptic feedback, multiplayer training
Phase 5 2027 Global expansion, LMS integrations, cognitive disability sims

💰 AWS Cost Estimate

Service Monthly (Low Usage) Monthly (10K Users)
CloudFront ₹250–₹600 ₹2,000–₹5,000
S3 ₹80–₹200 ₹500–₹1,000
Lambda ~₹0 (free tier) ₹800–₹2,000
API Gateway ~₹80 ₹600–₹1,500
Rekognition ₹0 (free tier 5K/mo) ₹2,000–₹5,000
Bedrock ₹200–₹800 ₹3,000–₹8,000
Cognito Free (up to 50K MAU) Free
Total ~₹800–₹1,800/mo ~₹9,000–₹22,000/mo

Well within ₹8,000 ($136) AWS credit for hackathon phase.


📜 Compliance Standards Referenced

  • ADA 2010 — Americans with Disabilities Act Standards for Accessible Design
  • WCAG 2.1 — Web Content Accessibility Guidelines Level AA
  • RPWD Act 2016 — Rights of Persons with Disabilities Act (India)
  • NBC 2016 — National Building Code of India
  • ASTM F1637 — Standard Practice for Safe Walking Surfaces

👤 About

Kalim Sayyed — Full Stack Developer & Project Lead

Faheem Shiledar — Backend & Project Enhancer


🌐 Links

Resource URL
Live Demo https://d2d1ibzdtgm1nq.cloudfront.net
GitHub https://github.com/kalimx03/WalkInMyShoes_OG
AWS Region us-east-1 (N. Virginia)
Technical Article https://builder.aws.com/content/3AdIzUsC2O9YviSkGYNYWMPKTvg/building-walkinmyshoes-an-immersive-accessibility-awareness-platform-using-ai-and-interactive-simulations

📄 License

Proprietary — All Rights Reserved © 2026 WalkInMyShoes


WalkInMyShoes — Building a Bharat where everyone can participate. Powered by Amazon Bedrock, Amazon Rekognition, and 5 more AWS services.

About

AI-powered WebXR platform to experience disability firsthand — Visual, Hearing & Motor simulations + AR Accessibility Auditor, deployed on AWS.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages