MaargDarshan • UTTAM

Navigation Reimagined
for Every Person

A precision spatial-awareness platform that connects visually impaired users with their caregivers in real time — powered by Edge AI, IoT sensors, and a design philosophy rooted in human dignity.

360° Spatial Radar

Four ultrasonic and Time-of-Flight sensors map the user's surroundings in every direction — delivering obstacle alerts with millimetre precision.

Edge AI Processing

An onboard ESP32-CAM runs a real-time AI pipeline with no cloud dependency — instant, localised object recognition that works even without an internet connection.

Haptic Feedback Engine

Directional vibration pulses encode both the proximity and bearing of obstacles — giving the user clear spatial information through touch alone, without breaking stride.

Assistive Technology • Built in India

About Maarg Darshan Care

A wearable AI-powered cap that restores spatial awareness for visually impaired individuals — enabling independence, confident movement, and safety-focused caregiving, every single day.

360°
Spatial Coverage
<50ms
Alert Latency
4
Sensor Directions
24/7
Live Monitoring
1-tap
Emergency SOS
About the Project
What Maarg Darshan Care is — and who it was built for

Maarg Darshan Care is a next-generation assistive technology platform built to restore independence for individuals living with visual impairments. At its core is a wearable smart cap embedded with ultrasonic transducers, Time-of-Flight modules, and an ESP32-CAM — all working in harmony to build a continuous, real-time 360° picture of the space around the user.

The platform connects the wearer directly to their caregivers through a live web dashboard displaying sensor telemetry, a camera feed, GPS coordinates, and system alerts — giving families and support workers the visibility and reassurance they need, from wherever they are.

What We Do
Six core capabilities that work together as a single, seamless system

360° Obstacle Detection

Four ultrasonic sensors sweep all directions at once, building a continuous spatial picture with colour-coded severity levels — updated in real time.

Live Camera Streaming

The onboard ESP32-CAM streams video directly to the caregiver dashboard with zero cloud dependency.

Haptic Feedback Engine

A precision vibration motor translates obstacle data into directional pulses so users understand their surroundings through touch alone.

Live GPS Tracking

Real-time coordinates and a visual path trace appear on the caregiver dashboard the moment the user starts moving.

Voice Guidance

Spoken alerts are delivered through the caregiver's device using the Web Speech API — clear, natural Indian English.

Emergency SOS

A single button press sends an immediate alert to registered contacts and emergency services with the user's live GPS coordinates.

System Architecture
How our decentralized digital twin ecosystem connects hardware, AI, and cloud in real-time
Smart Cap Arduino + ESP32
SERIAL UART
+ Wi-Fi MJPEG
User's Phone AI + Voice Engine
HTTPS API + SMS Bi-Directional Telemetry & SOS
Cloud Relay FastAPI Backend
WEBSOCKET Alerts & Remote Overrides
Guardian App SOS Receiver
REST API
Live Dashboard
Web Console

The Hardware Node

Zero-latency collision avoidance engine isolated from network drops. Features a dual-microcontroller design (Arduino + ESP32) to guarantee real-time physical safety while simultaneously maintaining high-bandwidth live video streaming.

The User Node

An invisible, hands-free auditory interface anchored by the "Hey Vision" wake-word. Executes RAM-defiant background processing and Bluetooth SCO routing overrides so the visually impaired user is never disconnected from the AI.

The Cloud Node

The central synchronization hub. Cryptographically pairs the user to their caretaker, securely routing high-frequency spatial telemetry, SOS interrupts, and real-time AI contextual processing at lightspeed.

The Guardian App

Engineered for aggressive emergency interception. Uses system-level BroadcastReceivers to pierce lock screens, maximize volume, and trigger sirens — even if the app is killed by the phone's operating system.

The Web Console

The high-fidelity command center. Dynamically renders a live 3D WebGL mirror of the user's hardware, provides continuous MJPEG camera streaming, and tracks real-time cartographic GPS coordinates.

Our Core Purpose

Our Mission for a Better World

We believe that visual impairment should never be a barrier to independence, safety, or living a confident and fulfilling life.

Independence and Confidence
Helping visually impaired people navigate safely

Every step taken outside should be a step taken with confidence. Our primary mission is to replace fear with freedom. By providing a reliable, 360-degree awareness of their surroundings, we empower visually impaired individuals to step out, explore, and navigate their daily lives without constantly relying on physical guidance from others.

Driven by Smart Technology
Combining Edge AI, precise sensors, and smart cameras

We harness the power of modern innovation to solve real-world challenges. By integrating ultrasonic sensors, advanced Time-of-Flight lasers, and Edge AI processing through the ESP32-CAM, we transform a simple wearable cap into a powerful "digital eye."

Real-Time Caregiver Support
Live tracking, video streaming, and instant alerts

True safety involves a support system. Our live dashboard provides caretakers with real-time GPS tracking, a live camera feed, and instant emergency SOS alerts. We bridge the distance between users and their loved ones, providing unparalleled peace of mind.

Affordable and Accessible
Democratizing safety for every socioeconomic background

Advanced assistive technology is often treated as a luxury, priced far beyond the reach of those who need it most. We are on a mission to democratize mobility by providing enterprise-grade safety at a fraction of the traditional cost.

Constant Evolution

Research & Innovation

Maarg Darshan Care is constantly evolving. By continuously researching and integrating smart technology, we aim to build the most advanced mobility aids for the visually impaired.

Our Research Areas
Core domains driving our intelligent platform

Smart Navigation Systems

Developing intelligent pathfinding algorithms that adapt to dynamic environments in real-time.

AI-Based Obstacle Detection

Enhancing machine vision to accurately classify complex obstacles using edge computing.

Wearable Assistive Tech

Designing ergonomic, discreet wearables that seamlessly integrate into the user's daily wardrobe.

Real-Time Monitoring

Creating zero-latency telemetry pipelines to keep caregivers connected without interruption.

Innovation Process
How we turn ideas into life-changing realities
1

Idea

Identifying core mobility challenges and brainstorming smart, accessible solutions.

2

Design

Drafting hardware schematics and designing intuitive user interfaces for maximum accessibility.

3

Prototype

Building functional hardware models to test sensor fusion and edge processing capabilities.

4

Testing

Conducting extensive real-world trials to ensure safety, reliability, and precision.

5

Deployment

Rolling out the finalized product and actively gathering user feedback for continuous improvement.

Technologies Used
The hardware powering our vision

Arduino

Manages low-level sensor timing and rapid component integration seamlessly.

ESP32 / ESP CAM

Delivers powerful edge AI processing and live video streaming over Wi-Fi.

Ultrasonic Sensors

Provides reliable 360° proximity data by measuring sound wave reflections.

TOF Sensors

Time-of-Flight lasers offer millimeter-precise distance measurements.

GPS Modules

Tracks live coordinates to ensure users and caregivers always stay connected.

Future Roadmap
Where we are heading next

Short Term Goals

Refining sensor accuracy, optimizing battery life, and expanding our real-world testing pool.

Mid Term Goals

Integrating advanced machine learning models for precise object classification like stairs or vehicles.

Long Term Vision

Building a global ecosystem of smart wearables that makes universal accessibility an affordable standard.

Device Guide

Device Guide

A quick setup and operation guide for your Maarg Darshan device.

1. Powering On & Hardware Setup

Press and hold the main recessed power button on the Smart Cap for 3 seconds. The localized bone conduction transducers will vibrate once, and the LED indicator will pulse blue, signaling that the dual-microcontroller engine is active and ready for pairing.

2. Pairing the User Application

Open the Maarg Darshan mobile app. The application is completely hands-free and voice-guided. It will automatically scan and pair with the cap via Bluetooth. Once connected, the app routes all AI audio directly through the cap's bone conduction hardware to keep your ears unoccluded.

3. Caretaker Cryptographic Linking

The caretaker must download the separate Caretaker App or log into the Web Dashboard. For absolute privacy, open searches are disabled. The caretaker must enter the secure, pre-generated pairing code generated by the User App to establish a cryptographically linked pipeline.

4. Testing the SOS Pipeline

Test the connection by pressing the recessed physical SOS button on the cap. The Caretaker App will instantly intercept the payload, forcing a red lock-screen override and sounding a maximum-volume siren on their phone. The caretaker can then press "Acknowledge" to send a voice confirmation back to you.

5. Initializing the Spatial AI

With setup complete, simply say "Hey Vision, what is around me?". The secondary microcontroller will capture an optical frame and route it to the cloud, returning a vivid spatial description of your environment within milliseconds.

The Minds Behind The Innovation

Meet Our Team

A dedicated group of engineers, designers, and innovators united by a single vision: making independent mobility safe, affordable, and accessible for everyone.

Debjeet Mazumder

Debjeet Mazumder

Hardware Designer / Software Developer

JIS College of Engineering

Leads hardware design, embedded systems logic, leads testing, ai development and overall system architecture.

Mehul Kumar

Mehul Kumar Jaiswal

AI & App Developer

JIS College of Engineering

Handles machine learning models and optimizes ultrasonic proximity algorithms.

Debadrita Baksi

Debadrita Baksi

Frontend Web Developer & UI/UX Designer

JIS College of Engineering

Architects the seamless fusion of intuitive UI/UX design, responsive frontend interfaces, and robust backend system integration.

Why Our Team is Special
The principles driving our collaborative success

Innovation First

Constantly exploring new edge-computing and AI methodologies to improve device latency and accuracy.

Dedicated Empathy

Designing with deep empathy for the end-user, ensuring technology solves real, human problems seamlessly.

Tech Mastery

Leveraging modern stacks from embedded C to React to deliver enterprise-grade stability on accessible hardware.

Need Help?

Frequently Asked Questions

Find answers to common questions about the Maarg Darshan system.

Does the cap require a constant internet connection to work?

No. The core obstacle detection runs locally on the edge using the ESP32 microcontroller, ensuring immediate haptic feedback even without Wi-Fi. However, live camera streaming and GPS tracking require a connection to sync with the caregiver dashboard.

How does the caregiver receive an SOS alert?

When an SOS is triggered, the system instantly pushes a high-priority alert to the web dashboard, sounding an alarm and logging the exact GPS coordinates.

How long does the battery last on a single charge?

The system is optimized for continuous daily use, providing reliable battery life to ensure the user remains connected and safe throughout their daily routines.

Can the haptic feedback be adjusted?

The vibration intensity scales dynamically based on how close the obstacle is—stronger vibrations mean the object is closer, providing an intuitive sense of space.

MaargDarshan feather

Welcome Back

Enter your credentials to access the dashboard.

Forgot?

Step 1 of 2 — Your Details

MaargDarshan feather

Create Account

Register your device to start monitoring.

Please enter a valid 10-digit mobile number.
Please enter a valid email address.
Passwords do not match.

Step 2 of 2 — Verify Your Identity

Verify Your Account

Enter the 6-digit OTP sent to your mobile and email address.

DEMO MODE — OTP shown for testing

——

Code expires in 02:00  · 

Didn't receive the code? Edit details

Documentation

MaargDarshan User Manual

Comprehensive guide to the hardware anatomy, navigation protocols, and the AI digital twin ecosystem.

Chapter 1: Hardware Anatomy & Ergonomics

The Cap

The physical assembly utilizes a counterbalanced mechanical structure. The frontal rigid tubular housing containing the directional sensors is perfectly counterweighted by a rear-mounted cotton spindle structure to prevent cervical strain on the user's neck.

The Enclosure

The entire apparatus is enclosed within a functional, breathable zipper-based fabric matrix. This custom zipper-enclosure conceals and protects the delicate hand-soldered internal electronics from the elements while featuring calculated incisions to allow ultrasonic waves and heat dissipation to pass through uninterrupted.

Bone Conduction Audio

The cap features integrated bone conduction transducers embedded into the lateral perimeter of the headgear chassis. These rest flush against the temporal or zygomatic bones, transmitting AI-generated auditory feedback and emergency confirmations directly through the cranial structure. This ensures the user's ear canals remain completely unoccluded, allowing them to simultaneously hear critical ambient auditory cues like approaching traffic.

Power, Charging, and Battery Architecture

The cap operates on a unified, highly optimized single-battery power architecture that sustains both microcontrollers, the haptic motors, and the camera simultaneously. Users must utilize the integrated localized charging module to recharge the device.

Chapter 2: Setup & Cryptographic Pairing

User App Initialization

The application operates hands-free and utilizes a Bluetooth Synchronous Connection-Oriented (SCO) routing override to pair with peripheral wearable microphones, ensuring high-fidelity audio capture even in dynamic outdoor environments.

Caretaker Linking

To guarantee absolute data privacy, the system does not allow open searches for users. During registration, the caretaker must input a specific, pre-generated secure pairing code or the exact registered email of the visually impaired user. This strictly authenticates and establishes a secure, cryptographically linked digital twin pipeline.

Chapter 3: Tactile Navigation & Spatial Awareness

The 3-Zone Haptic Engine

The hardware translates object proximity into three strict operational spatial zones using dynamically mapped Pulse Width Modulation (PWM): Awareness Zone (Far) triggers a low-intensity, slow heartbeat pulse. Caution Zone (Mid) acts as an intermediate warning zone. Danger Zone (Close) triggers a maximum-PWM, rapid-firing vibration localized precisely to the direction of the impending impact to instinctively prompt evasion.

5-Tier Object Classification System

The cap actively classifies the physical environment: Static Objects (distance remains constant), Objects in Motion (distance fluctuates independently), Elevations/Drop-offs (detected via ToF laser discrepancy), Walls/Flat Barriers (uniform acoustic reflections), and Standard Frontal Obstacles.

Tilt-Muting Protocol

An Inertial Measurement Unit (IMU) is mounted at a precise 90-degree transverse offset relative to the user's forward-facing axis. If the user intentionally tilts their head downward beyond a predefined threshold (e.g., >60 degrees), the logic board autonomously mutes the forward ultrasonic sensors to prevent false-positive "Danger" haptic responses.

Chapter 4: Advanced Movement Protocols

Walking/Corridor Mode (Triple-Tap)

In highly chaotic environments, the user can activate a dynamic threat-level enumeration by executing a deliberate tactile triple-tap gesture. This initiates a "Haptic Ducking" protocol, instantaneously attenuating the intensity of the left, right, and rear peripheral motors by 50% to 60%.

Fail-Safe Overrides

The cap includes automated safety protocols that bypass ducking suppression. Drift Detection triggers an isolated corrective pulse if the user's trajectory drifts perilously close to a lateral boundary. Threat Override instantaneously blasts maximum intensity to prevent an imminent collision if an active object moves rapidly toward the user's peripheral vector.

Chapter 5: The "Hey Vision" AI Assistant

Hardware Requirement for Voice Commands

To guarantee high-fidelity audio capture in noisy outdoor environments, the app executes a Bluetooth Synchronous Connection-Oriented (SCO) routing override. The user must wear a paired peripheral microphone because the smartphone's native internal microphone is bypassed when sitting in a pocket.

Pocket Mode Efficiency

To conserve battery and prevent accidental inputs, the application continuously polls the smartphone's ambient light and proximity sensors. If the device is placed in a pocket, it autonomously drops screen brightness to absolute zero and disables touch inputs while background voice Coroutines remain continuously active.

Spatial Queries

When the user asks a spatial query, the secondary microcontroller captures a high-resolution optical frame. The multimodal Large Language Model (LLM) is constrained by advanced prompt engineering to rapidly return highly specific, structured spatial data, detailing only the object's Color, Relative Size, and Exact Spatial Position.

Chapter 6: Emergency Protocols & Autonomous Recovery

Tri-Vector SOS Trigger

An emergency protocol can be initiated via a physical recessed button on the wearable apparatus, an acoustic distress voice command processed by the AI, or autonomous fall detection registered by sudden accelerometric and gyroscopic spikes in the IMU.

The Double-Tap Cancellation

Following an SOS fall state, the IMU continuously polls for vertical elevation indicating the user has stood up ("Rise Detection"). The SOS state is strictly neutralized only when the user executes a deliberate physical "double-tap" gesture on the wearable chassis. This final payload actively silences the sirens across the Caretaker Mobile App and Web Dashboard.

Chapter 7: Caretaker Operations & Oversights

RAM-Defiant Alerts

On the caretaker's mobile device, a high-priority BroadcastReceiver intercepts encrypted SOS SMS payloads even if the app is killed from active RAM. It forcefully acquires a CPU WakeLock, maximizes audio channels to loop a harsh siren, and paints a full-screen, un-dismissible red visual alert overlay.

The Two-Way Handshake

When the caretaker receives the alert, they are presented with an "Acknowledge" interface. Pressing this transmits a priority network payload back to the visually impaired user's smartphone, announcing through the wearable speakers: "Help is on the way".

The AI Scan History

The Caretaker Web Dashboard integrates a synchronized AI Scan History ledger. Caretakers can asynchronously review a scrollable feed containing the exact high-resolution optical frame captured by the user's hardware, the voice-transcribed question, and the specific spatial answer generated by the multimodal AI.

Chapter 8: The Caretaker Web Dashboard (The Command Center)

3D Digital Twin (WebGL Mirroring)

The web dashboard renders a live, interactive 3D model of the user's physical smart cap, illuminating corresponding quadrants (glowing red or yellow) as physical obstacles approach the user, providing instant visual context.

Live Sensor & Movement Telemetry

The dashboard features dedicated status panels where caretakers can monitor real-time numerical distance data (in centimeters) for all four directions, alongside vibration motor status, IMU tilt level, buzzer activity, and battery life.

Live Camera Streaming & GPS Tracking

The dashboard connects directly to the cap's ESP32-CAM to render a continuous, zero-cloud-dependency MJPEG live video feed, while simultaneously dropping a real-time tracking marker on a live cartographic map.

Browser-Based Emergency Takeover

If the user triggers an SOS, the website actively intercepts the server payload and instantly overrides the standard interface with a flashing, full-screen red visual alert and maximum-volume HTML5 emergency siren.

Chapter 9: Asynchronous Caretaker Monitoring

The AI Scan History Ledger

Whenever the user initiates a "Hey Vision" spatial query to the AI, the dashboard fetches the payload and populates a scrollable feed. Caretakers can asynchronously review the exact high-resolution optical frame captured by the hardware, the transcribed question asked by the user, and the AI's exact spatial response, allowing for a complete audit of the user's daily interactions.

Chapter 10: Caretaker Mobile Application Architecture

Secure Cryptographic Account Pairing

The app explicitly disables open directory searches for users. During registration, the caretaker must enter a specific, secure pre-generated pairing code or the user's exact registered email to authenticate and securely mirror the user's data stream.

Synchronized AI Scan Feed

Whenever the visually impaired user makes a "Hey Vision" voice request, the caretaker's mobile app receives a synchronized real-time payload, updating a scrollable ledger containing the base64 image, user question, and AI spatial response.

RAM-Defiant System Interception

The Caretaker App deploys a high-priority system BroadcastReceiver operating continuously at the kernel level. Even if the Caretaker App has been completely wiped from active RAM, this receiver intercepts incoming emergency payloads instantly.

Lock-Screen Red Alert Override

Upon intercepting a crisis signal, the BroadcastReceiver acquires a system CPU WakeLock, overrides the phone's native audio state (bypassing "Silent" or "Vibrate" profiles), and loops a high-decibel emergency siren with a full-screen red visual alert overlay directly over the lock screen.

Autonomous Safe-State Clearing

If the user executes a physical double-tap to signal an intentional fall-recovery post-impact, the caretaker app intercepts the "Safe State" payload. It immediately terminates the system CPU WakeLock, silences the maximum-volume sirens, replaces the red overlay with a green safety notice, and safely closes the emergency loop.

Directions
Front-- cm
Left-- cm
Right-- cm
Rear-- cm
Movements
Vibration MotorIDLE
IMU / TiltLEVEL
BuzzerOFF
Session
0
Alerts fired
00:00
Uptime
78%
WAITING FOR STREAM
ESP32-CAM
--:--:--
AI Ready
3D Sensor Visualization

Real-time physical mirror. Highlights indicate obstacle proximity based on active sensor data.

LIVE — Polling ESP32
360° Proximity Radar
GPS Location
Last Known Location Click to view live map
Waiting for GPS...
Event Log
--:--:--System initialised
AI Image Scan History
No scans yet