Skip to content
BranchesTags

Repository files navigation

🚜 FogBot — Safe & Efficient Operation of Mine Vehicles in Fog and Low-Visibility Conditions

SIH 2026 NMDC Problem Statement SIH26007 Target Site: Bailadila Iron Ore License: ISC Node.js Three.js Docker Compose

Smart India Hackathon 2026 — Project Report & Software Command Center
Problem Statement ID: SIH26007 | Organization: NMDC Limited
Team: Track Decoders | Repository: github.com/Arka-124/FogBot
Live Web Presence & Digital Twin: fogbot.onrender.com

📌 Table of Contents

  1. Executive Overview & Problem Statement
  2. Our Chosen Approach: Leading Pilot Rover — FogBot
  3. Prototype Hardware & Decided Sensor Stack
  4. Sensor Fusion & AI Decision Pipeline
  5. Software Architecture, Command Center & 3D Digital Twin
  6. Repository Structure
  7. Quickstart & Local Installation Guide
  8. Cloud Deployment Blueprints (Render & Railway)
  9. Milestone Roadmap & Feature Priority Tiers
  10. Judge Demo Narrative (7-Step Story Arc)
  11. Open Engineering Questions & Future Horizons
  12. Team & Acknowledgments

1. Executive Overview & Problem Statement

The Bailadila Monsoon Fog Crisis

NMDC’s Bailadila Iron Ore Complex (Kirandul and Bacheli Complexes in Dantewada, Chhattisgarh) is one of India's largest and highest-grade mechanized iron ore deposits. However, during the prolonged monsoon season, dense hilltop clouds and heavy fog settle across the opencast mines, dropping visibility to a critical 3–5 meters on narrow, winding hilltop haul roads.

This environmental bottleneck costs the complex 50–60 operational days per year, creating:

  • Catastrophic Collision Risks: High-capacity Heavy Earth Moving Machinery (HEMM) dumpers operating within blind stopping distances of other dumpers, light utility vehicles (LUVs), water tankers, and maintenance personnel.
  • Forced Haulage Halts: Inability to safely navigate hilltop curves leading to complete stoppage of haul cycles.
  • Depressed Fleet Productivity & Ore Evacuation: Bailadila currently produces ~37 MTPA, with aggressive strategic expansion targets of 80 MT by 2030. Losing nearly two full months of haul operations per annum threatens national raw material supply chains.
  • Inadequacy of Traditional Aids: Conventional headlights, high-intensity fog lamps, reflective road markers, and basic microwave radars scatter or blind operators in zero-visibility conditions.

Documented Industry Hazards & Real-World Precedents

Incident / Mining Case Documented Finding / Impact Relevance to FogBot Architecture
Mine Haul Road Fatality Investigation (2010) Haul truck operator locked brakes on an unlit road before rear-ending a stationary lead vehicle. Haul trucks cannot stop instantaneously on mine grades; alerts must originate ahead of the vehicle.
NSW Open-Cut Near-Miss (2021) 240-tonne haul truck narrowly missed a light vehicle at an intersection due to blind spots. Validates necessity of autonomous active forward scanning and cross-vehicle safety broadcast.
Open-Pit Collision Statistics (2020–2023) 12% surge in truck-ancillary collisions; ~25% of collisions involve another heavy vehicle within 50 m. Proves that human line-of-sight in industrial environments is inherently flawed during degraded weather.
Vale’s Carajás Complex Weather Halts Mine halted production across rainy/foggy shifts; output fell 19.1% QoQ due to weather stoppages. Highlights that global mining leaders suffer identical bottlenecks without leading guidance.
Rio Tinto Pilbara Weather Disruption Disrupted evacuation lost ~13 million tonnes in a single quarter. Direct fiscal proof that solving 50–60 lost days unlocks massive enterprise ROI for NMDC.

Why Existing Aids Fail: Critique of Bailadila's Existing Smart DAS

A field-tested Smart Driving Assistance System (DAS) is already deployed at Bailadila's Bacheli Complex. It integrates:

  1. Differential GNSS tracking
  2. Proximity radars mounted on truck bumpers
  3. Anti-collision forward laser projection lines
  4. Electrically heated road-edge guidance lighting

The Crucial Insight: Despite this installed infrastructure, the site still loses 50–60 days every single year.

  • The Truck-Mounted Sensor Fallacy: Mounting sensors directly on an 85- to 240-tonne HEMM truck means the sensors are subject to heavy chassis vibration, immense dust accumulation, and are physically situated inside the same dense fog bank that blinds the driver. If an obstacle is detected 8 meters ahead of a loaded dumper traveling at haul speed on a 1:16 gradient, physical braking distance exceeds the detection distance.
  • FogBot’s Paradigm Shift: Rather than trying to enhance the truck's degraded cockpit perspective, FogBot decouples sensing from hauling. A dedicated, agile pilot rover travels ahead of the truck, penetrating the fog first, establishing a safe buffer, and feeding processed trajectory intelligence back to the trailing driver.

2. Our Chosen Approach: Leading Pilot Rover — FogBot

Instead of retrofitting capital-intensive sensor suites onto hundreds of individual haul trucks, our team is building FogBot — an autonomous, purpose-engineered pilot rover that precedes the haul truck, holding a dynamic safety distance and acting as the vehicle’s forward eyes and scout.

[ HAUL ROAD PATH ]
========================================================================================
     [ FogBot Pilot Rover ]  <--- Dynamic Distance (15-25m) --->  [ HEMM Dumper Truck ]
     • 360° STL-19P LiDAR                                          • Operator Display
     • CV Camera + 3D Depth                                        • High-Decibel Siren
     • Ultrasonic + IR Arrays                                      • Linked E-Stop Trigger
     • AI Collision Risk Engine                                    • Following Safe Speed
========================================================================================

Engineering Risk vs. Mitigation Matrix

Engineering Challenge / Risk Severity Root Cause FogBot Architectural Mitigation
Rover navigating zero visibility High Rover faces the same dense fog as the truck. Rover carries the full multi-sensor fusion stack (LiDAR + CV Camera + IR Proximity + Ultrasonic + GNSS), computing safety scores at ground level.
Stalled / Crashed Rover hazard Critical A disabled rover directly in front of a loaded 100t dumper is an immediate collision hazard. Mandatory Fail-Safe Protocol: Dual-redundant heartbeat. If rover loses localization, battery, or motor heartbeat, an immediate broadcast triggers auto pull-off or instantaneous hard-stop alert to the trailing truck.
Constant-distance holding Medium Commercial GPS suffers from 3–5m drift in hilly terrain. Rover fuses GPS absolute position with real-time wheel odometry and close-range IR/ultrasonic telemetry; planned UWB ranging module for sub-meter gap precision.
Rover-to-Truck coordination complexity Medium Dynamic telemetry transmission in remote pit conditions. Low-latency direct RF / Wi-Fi mesh telemetry broadcast backed by standard MQTT/WebSocket publish-subscribe architecture.

3. Prototype Hardware & Decided Sensor Stack

Physical Prototype Hardware Specs

Our functional proof-of-concept prototype is built on an industrial MentorPi-style robotics platform engineered for rugged embedded perception:

  • Primary Edge Compute: Raspberry Pi 5 (4GB RAM) running 64-bit Linux OS for computer vision inference, sensor fusion processing, and telemetry broadcasting.
  • Low-Level Motion Control: STM32 ARM Cortex microcontroller handling closed-loop PID motor control, encoder feedback, ultrasonic/IR interrupts, and emergency stop relays.
  • Actuation & Drive: High-performance metal gear DC encoder motors providing precise odometry feedback.
  • Power Delivery: High-discharge LiFePO4 / Li-ion multi-cell power regulation with isolated logic and motor power rails.

Final Multi-Sensor Stack

+-----------------------------------------------------------------------------------+
|                            FOGBOT PERCEPTION SUITE                                |
+------------------------------------+----------------------------------------------+
| SENSOR                             | PRIMARY OPERATIONAL FUNCTION                 |
+------------------------------------+----------------------------------------------+
| STL-19P TOF LiDAR                  | Medium/long-range 360° point-cloud mapping,  |
|                                    | SLAM localization, haul road edge detection   |
| Vision Camera + 3D Depth Camera    | Real-time obstacle classification (dumpers,  |
|                                    | boulders, personnel) & depth point mapping    |
| IR Distance Sensors                | High-speed close-range obstacle confirmation  |
| Ultrasonic Sensor Array            | Complementary acoustic ranging (penetrates    |
|                                    | dark, glossy, and reflective mud surfaces)    |
| GNSS / GPS Module                  | Haul road waypoint tracking and map spatial   |
|                                    | coordinates                                  |
+------------------------------------+----------------------------------------------+

Deliberate Hardware Omissions & Fog Proxy Innovation

  1. Explicitly Excluded Hardware:
    • ❌ Thermal Imaging Cameras: Rejected due to high thermal noise and blooming caused by rain droplets and saturated moisture in tropical monsoon clouds.
    • ❌ mmWave Radar: Omitted from current build to avoid multi-path scattering against wet, steep iron ore bench faces, and to drastically lower hardware bill of materials (BOM).
  2. The CV Detection-Confidence "Fog Proxy": Rather than purchasing dedicated, expensive optical scatterometers or transmissometers, FogBot turns its optical camera into a real-time fog meter: $$\text{Fog Density Proxy} \propto 1.0 - \text{Confidence}_{\text{CV Model}}(\text{Known Markers / Terrain})$$ As water droplets attenuate light transmission, convolutional feature confidence decays predictably. When camera confidence drops, the AI safety engine dynamically shifts sensor weighting toward LiDAR and ultrasonic readings, while recalculating safe stopping distances.

4. Sensor Fusion & AI Decision Pipeline

Architecture Flowchart

flowchart TD
    subgraph SENSING["Perception Layer"]
        CAM["Vision & 3D Depth Camera"]
        LIDAR["STL-19P TOF LiDAR"]
        IR["IR Proximity Sensors"]
        US["Ultrasonic Array"]
        GPS["Outdoor GNSS / GPS"]
    end

    subgraph PROCESSING["Edge Processing & Feature Extraction"]
        CV_CONF["Object Detection & Confidence Score<br/><i>(Fog-Density Live Proxy)</i>"]
        FUSION["Multi-Sensor Fusion Engine<br/><i>(Complementary Cross-Checking)</i>"]
    end

    subgraph RISK_ENGINE["AI Safety & Risk Engine"]
        CALC["Compute Continuous Collision Risk Score<br/>Inputs: Distance + Closing Speed + Fog Proxy + Curvature"]
    end

    subgraph DECISION["Adaptive Execution States"]
        SAFE["SAFE DRIVE<br/>Normal Haul Speed"]
        CAUTION["CAUTION SLOW<br/>Reduced Speed + Alert"]
        STOP["EMERGENCY STOP<br/>Auto-Brake + Instant Broadcast"]
    end

    subgraph ACTUATION["Vehicle & Fleet Response"]
        STM32["STM32 Motor Controller"]
        TRUCK_LINK["Rover-to-Truck Broadcast Link<br/><i>(TTC Alarm to Operator)</i>"]
        CLOUD["IoT Central Dashboard<br/><i>(MQTT / WebSockets)</i>"]
    end

    CAM --> CV_CONF
    CV_CONF --> FUSION
    LIDAR --> FUSION
    IR --> FUSION
    US --> FUSION
    GPS --> FUSION

    FUSION --> CALC
    CALC -->|Risk = LOW| SAFE
    CALC -->|Risk = MEDIUM| CAUTION
    CALC -->|Risk = HIGH / Critical| STOP

    SAFE --> STM32
    CAUTION --> STM32
    STOP --> STM32

    CALC --> TRUCK_LINK
    CALC --> CLOUD
Loading

Collision Risk Scoring Engine

Rather than relying on basic binary obstacle flags ("obstacle detected: yes/no"), FogBot's embedded safety core executes continuous multi-parameter risk evaluation: $$\text{Risk Score} = f\Big(\Delta d_{\text{rover-truck}},, v_{\text{closing}},, (1 - C_{\text{cam}}),, d_{\text{obstacle}},, \kappa_{\text{road}}\Big)$$

  • If camera confidence $C_{\text{cam}}$ drops, the system classifies environmental fog as severe and contracts safe threshold buffers.
  • If distance to obstacle $d_{\text{obstacle}}$ falls below the physical stopping envelope for current speed, risk escalates instantly to HIGH.

Dynamic Adaptive Speed Logic

Camera Confidence $C_{\text{cam}}$ (Fog Proxy) Haul Road Visibility AI System Behavior & Actuation Recommended Speed
High (> 75%) Clear / High (> 50m) Normal autonomous pilot mode; full LiDAR SLAM path execution. 15–20 km/h
Moderate Drop (40%–75%) Light Fog (20m–50m) Proactive deceleration; sensor fusion shifts higher weight to LiDAR. 10–14 km/h
Sharp Drop (15%–40%) Dense Fog (5m–20m) Crawl speed; heightened ultrasonic and IR close-range hazard polling. 5–8 km/h
Near-Zero (< 15%) Critical (3m–5m) Emergency Hold/Stop: flashing rear beacon, siren, halt truck before hazard. 0 km/h (HOLD)

Rover ↔ Haul Truck Link & Mandatory Fail-Safe Stop

The pilot rover coordinates with its assigned trailing HEMM dumper via an ultra-reliable wireless safety channel:

  1. Continuous Distance-Hold Control Loop: The rover broadcasts its real-time GPS coordinates, velocity, and computed safety envelope. The dumper cab receiver displays distance gap, relative speed, and calculated Time-to-Collision (TTC).
  2. Immediate Fail-Safe Stop Broadcast: In an open-pit environment, if the lead rover suffers a mechanical breakdown, battery failure, or lost localization, it becomes an unlit stationary obstacle. FogBot implements an active heartbeat watchdog:
    • If rover health heartbeat ceases or an emergency halt is triggered, an instantaneous E-STOP packet floods the frequency band.
    • The truck cabin unit sounds an immediate high-decibel audible alarm and illuminates visual brake warnings, halting the dumper well before closing the gap.

5. Software Architecture, Command Center & 3D Digital Twin

Two-Track Agile Strategy (Hardware ‖ Software)

To guarantee complete demo readiness for hackathon judging without waiting for physical hardware assembly:

  • Track 1 (Hardware & Robotics): Chassis assembly $\rightarrow$ STM32 bring-up $\rightarrow$ LiDAR/Camera fusion $\rightarrow$ Field trials.
  • Track 2 (Software, Dashboard & Digital Twin): Landing page $\rightarrow$ Telemetry JSON schema $\rightarrow$ Mock data stream $\rightarrow$ 3D Digital Twin $\rightarrow$ Integration merge.
Hardware Track:  Sensors ---> Sensor Fusion ---> AI Risk Engine ---> Live MQTT Telemetry Stream ---\
                                                                                                    ===> Unified SIH Demo
Software Track:  Mock Generator ---> Command Center UI ---> 3D WebGL Digital Twin ----------------/

Web Presence & Live Condition Monitor (index.html)

The root web portal (http://localhost:3000/) serves a single-scroll command interface designed with an industrial dark theme (Slate-950 and Rajdhani typography):

  • Live Bailadila Condition Ticker: Real-time atmospheric visibility metric (e.g. 86% visibility) updated via live polling.
  • Problem Statement KPI Cards:
    • 3–5m worst-case monsoon visibility
    • 60 days/yr lost mining operations
    • 37 MTPA current Bailadila iron ore output
    • 80 MT NMDC 2030 strategic vision
  • Direct Operator Portal Access: Single-click navigation to the secure authenticated control center.

Interactive 3D Digital Twin (rover3d.bundle.js / Three.js)

FogBot includes an interactive 3D WebGL Digital Twin built directly with Three.js (rover3d.js bundled via esbuild to rover3d.bundle.js):

  • Industrial Rover Geometry: Modeled chassis, articulated high-traction wheels, top-mounted STL-19P LiDAR tower, front-facing dual camera rig, and rear hazard assembly.
  • Dynamic Exponential Volumetric Fog: Three.js FogExp2 shader whose density dynamically expands and contracts based on incoming telemetry: $$\text{density} = \max(0.005, \min(0.025, 0.04 - \text{visibility} \times 0.0004))$$
  • Pulsing Rear Hazard Beacon: PointLight strobe simulating high-intensity warning beacons required on mine haul roads.
  • Color-Coded Status Materials: Real-time switching between SAFE (Cyan/Emerald), CAUTION (Amber), and HIGH / E-STOP (Crimson Red).
  • Interactive OrbitControls: Full damping-assisted orbit, pan, and zoom for judge inspection.

Production-Grade Operator Login Gateway (login.html & server.js)

Access to the fleet dispatch command center is safeguarded by an enterprise authentication gateway:

  • Google reCAPTCHA v2: Mandatory client-side challenge and strict server-side verification before database query execution to block automated brute-force attacks.
  • Parameterized MySQL / MariaDB Queries: Prepared statements executed through mysql2/promise pool to ensure zero SQL injection vulnerability.
  • Bcrypt Password Security: Passwords hashed and compared with salts via bcryptjs.
  • Resilient Cloud In-Memory Fallback: When running in demo environments without persistent MySQL (e.g., Render free web service), the backend automatically engages an in-memory credential store with full logging transparency.

Central IoT Telemetry Architecture & JSON Schema

The telemetry pipeline communicates over MQTT (Mosquitto broker) and WebSockets. The JSON schema standardizes communication between physical rovers, simulated test harnesses, and the frontend:

{
  "rover_id": "ROVER_01",
  "timestamp": "2026-09-11T10:21:08Z",
  "gps": {
    "lat": 18.672500,
    "lon": 81.324700
  },
  "speed_kmh": 12.0,
  "fog_visibility_m": 18,
  "camera_confidence_pct": 31,
  "lidar_confidence_pct": 88,
  "obstacle": {
    "detected": true,
    "type": "human",
    "distance_m": 6.2,
    "source": "ir_ultrasonic_cross_check"
  },
  "risk_level": "HIGH",
  "recommended_speed_kmh": 5,
  "gap_to_truck_m": 18.2,
  "time_to_collision_s": 2.1,
  "status": "CAUTION"
}

Judge-Facing Controls: Live Fog-Density Slider & E-Stop

Designed specifically for live SIH demonstration:

  1. Interactive Fog-Density Slider: Judges can drag a slider from 100 m down to 3 m visibility. As the slider moves:
    • Camera confidence visibly drops.
    • Sensor weight shifts in real time toward LiDAR.
    • Recommended speed throttles automatically down to 0 km/h.
    • 3D Digital Twin volumetric fog thickens in real time.
  2. One-Touch E-Stop Button: Instantly broadcasts an emergency stop payload across the network, demonstrating fail-safe interlock.

6. Repository Structure

FogBot/
├── assets/                       # Visual assets, branding, and imagery
│   ├── nmdc-logo.png             # Official NMDC emblem
│   ├── sih-logo.png              # Smart India Hackathon 2026 logo
│   ├── rover-photo.png           # Hardware prototype reference photo
│   ├── rover.png                 # Pilot rover graphic & favicon
│   └── scan_points.json          # STL-19P LiDAR 1,500-point point cloud scan
├── css/                          # Modular stylesheets
│   ├── style.css                 # Base design system & login portal styles
│   ├── style-landing.css         # 3D Digital Twin landing page styles
│   ├── style-dashboard.css       # In-Cab HUD & Command Center layout
│   └── theme.css                 # Cyber-slate light/dark theme tokens
├── js/                           # Client-side scripts
│   ├── dashboard.js              # Command center telemetry, radar & HUD engine
│   ├── rover3d.bundle.js         # Minified standalone 3D twin bundle
│   ├── script-landing.js         # Landing page animations & telemetry counters
│   ├── script.js                 # Login portal auth & reCAPTCHA controller
│   └── theme.js                  # Zero-FOUC theme resolver & switcher
├── src/                          # Developer source & component modules
│   ├── rover3d.js                # Three.js 3D Digital Twin engine (ES Module source)
│   └── RoverDigitalTwin3D.jsx    # Standalone React Three Fiber component
├── database/                     # Database schemas & migrations
│   └── init.sql                  # MySQL 8.0 schema and seed user credentials
├── .agents/                      # Autonomous coding agent blueprints & skills
├── dashboard.html                # Pilot Command Center & In-Cab HUD interface
├── index.html                    # 3D Digital Twin landing page & system overview
├── login.html                    # Operator dispatch authentication interface
├── server.js                     # Express.js REST API & WebSocket real-time server
├── docker-compose.yml            # Containerized MySQL 8.0 service definition
├── package.json                  # Node.js project manifest and build scripts
├── package-lock.json             # Locked dependency tree
├── render.yaml                   # Infrastructure-as-code blueprint for Render
├── AGENTS.md                     # Project invariants & architectural rules
├── .env.example                  # Environment configuration template
├── .env                          # Local environment secrets (ignored by Git)
├── .gitignore                    # Version control exclusion rules
└── README.md                     # Comprehensive project engineering documentation

7. Quickstart & Local Installation Guide

Prerequisites

  • Node.js: v18.0.0 or higher (node -v)
  • npm: v9.0.0 or higher (npm -v)
  • Docker & Docker Compose (optional, for local containerized MySQL)

1. Clone the Repository

git clone https://github.com/Arka-124/FogBot.git
cd FogBot

2. Configure Environment Variables

Copy .env.example to .env:

cp .env.example .env

Ensure your .env contains:

PORT=3000
DB_HOST=localhost
DB_USER=root
DB_PASSWORD=your_mysql_password
DB_NAME=login_portal
DB_PORT=3306

# Google reCAPTCHA v2 Keys
RECAPTCHA_SITE_KEY=6LekebUtAAAAAEiqVaTTW15PdF-Z2ZH47YNGUalw
RECAPTCHA_SECRET_KEY=6LekebUtAAAAADmjFOUUelkHfkD8mSFyUPKGUCBL

3. Start Database (Optional via Docker)

If you wish to use the containerized MySQL 8.0 instance:

docker compose up -d

Note: If Docker or MySQL is not running locally, the server will automatically activate its internal resilient demo store.

4. Install Dependencies

npm install

5. Build 3D Digital Twin Bundle (Optional)

If modifying rover3d.js:

npm run build:3d

6. Start the Server

npm start

The server will bind to 0.0.0.0:3000:

Default Demo Credentials

Field Value Role
Operator User ID admin Lead Dispatcher / Safety Officer
Password password123 Authorized Shift Access

8. Cloud Deployment Blueprints (Render & Railway)

Option A: Render (Zero-Config Web Service)

This repository includes a native render.yaml blueprint:

  1. Connect your GitHub repository to render.com.
  2. Create a new Web Service pointing to FogBot.
  3. Configure settings:
    • Environment: Node
    • Build Command: npm install
    • Start Command: node server.js
  4. Set Environment Variables:
    • RECAPTCHA_SITE_KEY: 6LekebUtAAAAAEiqVaTTW15PdF-Z2ZH47YNGUalw
    • RECAPTCHA_SECRET_KEY: 6LekebUtAAAAADmjFOUUelkHfkD8mSFyUPKGUCBL
  5. Deploy service. Render will serve the application with automated HTTPS.

Option B: Railway (Node + Dedicated MySQL)

  1. In railway.app, click New Project $\rightarrow$ Deploy from GitHub repo.
  2. Add a MySQL plugin from the Railway service catalog.
  3. Railway automatically populates MYSQLHOST, MYSQLUSER, MYSQLPASSWORD, and MYSQLDATABASE.
  4. Add RECAPTCHA_SITE_KEY and RECAPTCHA_SECRET_KEY.
  5. The server.js startup lifecycle auto-creates the schema and demo user on first connection.

9. Milestone Roadmap & Feature Priority Tiers

Roadmap: September 2026 $\rightarrow$ Grand Finale December 2026

[Sep 6 - 21]      Phase 1: Research, concept lock, Bailadila DAS critique, architecture sign-off
[Sep 22 - Oct 12] Phase 2: Hardware chassis bring-up (Raspberry Pi 5 + STM32 + LiDAR + Cameras)
[Oct 13 - Nov 2]  Phase 3: Sensor fusion pipeline, AI risk engine, distance-hold control loop
[Nov 3 - 16]      Phase 4: Field integration testing (RC/scale trials, fog-generator chamber test)
[Nov 17 - 30]     Phase 5: Command center polish, 3D digital twin telemetry sync, technical paper
[Dec 1 - Finale]  Phase 6: Full rehearsal, judge defense prep, backup video verification

Feature Priority Tiers

+-----------------------------------------------------------------------------------+
| TIER 1: MUST BUILD (Core Rover + Closed-Loop Safety Loop)                        |
| • LiDAR + CV Camera + IR + Ultrasonic Sensor Fusion Pipeline                      |
| • YOLO / MobileNet Obstacle Classification & Detection Model                      |
| • Dynamic Collision Risk Engine (Camera confidence as live fog proxy)             |
| • Rover <-> Truck distance-holding control & instant E-Stop broadcast             |
| • Haul road GPS position tracking & Command Center dashboard                      |
+-----------------------------------------------------------------------------------+
| TIER 2: HIGH IMPACT (Operational Intelligence)                                    |
| • Adaptive Safe-Speed Regulation Algorithm                                        |
| • V2I (Vehicle-to-Infrastructure) warning beacons at critical haul-road turns      |
| • Production dispatch UI & real-time telemetry streaming                          |
+-----------------------------------------------------------------------------------+
| TIER 3: JUDGE "WOW" FACTOR (Advanced Simulation & Replay)                         |
| • Interactive 3D Digital Twin with WebGL shader-driven volumetric fog (Built)     |
| • Multi-vehicle fleet dispatch simulation                                         |
| • Black-box flight recorder & telemetry incident replay                           |
+-----------------------------------------------------------------------------------+
| TARGET PRODUCTION ARCHITECTURE (Described in Report, Excluded from Prototype)     |
| • PostGIS spatial databases & InfluxDB time-series clusters (cut to minimize ops) |
| • Multi-tenant enterprise SSO (simplified to operator role auth for hackathon)    |
+-----------------------------------------------------------------------------------+

10. Judge Demo Narrative (7-Step Story Arc)

When demonstrating FogBot to the Smart India Hackathon jury, our team presents a cohesive, end-to-end operational story arc:

  1. Phase 1 — Clear Haul Operations: Rover leads the dumper along the bench haul road at full authorized speed. Visibility is clear ($&gt;50\text{ m}$), camera confidence is high ($&gt;90%$), and system risk status is SAFE (Green).
  2. Phase 2 — Incursion of Monsoon Fog: The judge pulls the live Fog-Density Slider to $10\text{ m}$. In the software dashboard and 3D digital twin, camera confidence plummets. The fusion pipeline dynamically increases weighting on the STL-19P LiDAR and ultrasonic sensors.
  3. Phase 3 — Proactive Speed Adaptation: The AI Safety Engine calculates elevated risk; the recommended safe speed drops smoothly from $20\text{ km/h}$ to $8\text{ km/h}$, preventing forced haul halts while maintaining safe stopping distances.
  4. Phase 4 — Approaching Lead Hazard: An oncoming vehicle or stalled obstacle appears on the haul road. The distance-hold monitor calculates shrinking Time-to-Collision (TTC) and sounds a caution alert.
  5. Phase 5 — Multi-Sensor Confirmation & E-Stop: The obstacle crosses into the critical 6-meter envelope. Cross-checked IR and ultrasonic telemetry confirm positive detection. Risk escalates to CRITICAL (Red); the rover initiates emergency deceleration and broadcasts a hard-stop packet.
  6. Phase 6 — Dispatch Visibility: The Command Center dashboard updates within $&lt;50\text{ ms}$, alerting the central mine controller with exact GPS coordinates and sensor diagnostic readings.
  7. Phase 7 — 3D Digital Twin Replay: The WebGL 3D digital twin mirrors the rover's exact physical halt, flashing its rear warning beacon to safeguard the approaching haul truck.

11. Open Engineering Questions & Future Horizons

  • Sub-Meter Ranging Enhancement: While GPS provides reliable global positioning ($3\text{--}5\text{ m}$), tight haul road convoying requires sub-meter accuracy. Our roadmap integrates Ultra-Wideband (UWB) transceivers between the rover and trailing truck for centimeter-precise distance holding.
  • Haul Road Suspension & Incline Dynamics: Bailadila's haul roads feature rugged iron ore rubble and $1:16$ gradients. Testing will validate brushless motor torque ratios and track-wheel grip under slippery, red-mud monsoon conditions.
  • Inertial Measurement Unit (IMU) Integration: Integrating 6-axis IMU sensing into the STM32 board to detect rover chassis pitch/roll and trigger an automated rollover SOS broadcast.

👥 Team & Acknowledgments

  • Team: Track Decoders
  • Competition: Smart India Hackathon 2026 (SIH 2026)
  • Nodal Agency / Problem Creator: NMDC Limited (National Mineral Development Corporation)
  • Problem Statement: SIH26007 — Safe & Efficient Operation of Mine Vehicles in Fog / Low-Visibility Conditions
Engineered with precision for the mines of Bailadila. Built for Smart India Hackathon 2026.

About

Autonomous Leading Pilot Rover & IoT Command Center for Safe Operation of Mine Vehicles in Dense Fog | NMDC Problem Statement SIH26007 (SIH 2026)

fogbot.onrender.com

Topics

autonomous-roverbailadila-minedigital-twiniot-dashboardlidarmining-safetynmdcroboticssensor-fusionsih2026smart-india-hackathon-2026threejs

Resources

Readme
Activity

Stars

1 star

Watchers

0 watching

Forks

0 forks
Report repository

Releases

Packages

Contributors

Languages