Skip to content

WORD_DOC_TEMPLATE.md

Latest commit

 

History

History
40 lines (30 loc) · 3.63 KB

WORD_DOC_TEMPLATE.md

File metadata and controls

40 lines (30 loc) · 3.63 KB
Raw

Word Document Template for Unstop Submission

Instructions: Copy this content entirely into a Microsoft Word document (.docx) and upload it via the Unstop portal. It explicitly fulfills all textual requirements of the Rule Book (Section 1.3 and Section 7.1).

Team Name: [Enter Your Team Name]
Track Chosen: RoadSoS
Hackathon Theme: AI in Road Safety (CoERS, IIT Madras)

1. Introduction

Our submission, the Aapda Seva Command Center, tackles the RoadSoS track by providing a location-based platform that unifies all critical post-crash infrastructure. It directly impacts the "golden hour" by automatically aggregating coordinates and predicting AI-triage crash severity, eliminating human delay when routing immediate help to accident locations.

2. Comprehensive List of Software Packages Used

To satisfy the rulebook preference for open models, free APIs, and Python architectures:

AI Models & Backend (Python Preferred):

  • Python 3.10: Core processing environment.
  • Scikit-Learn & Pandas: Open-source analytical models handling the telemetry triage and sorting the highest severity crashes instantly.
  • JSON Libraries: Used to strictly format the submitted structured database for the AI model pipeline.

Application Interface & Frameworks:

  • React.js 18.x (Vite build system): Optimized to ensure zero-lag render times for the dispatcher dashboard.
  • Tailwind CSS & shadcn/ui: Component library guaranteeing strict visual contrast and accessibility.
  • date-fns & lucide-react: Core UI utility packages.

Information Integration & Open APIs:

  • OpenStreetMap (OSM) / Overpass API (Free/Open): Enables global applicability across countries, fetching towing, hospitals, showrooms, and puncture shops globally without proprietary mapping lock-ins.
  • Firebase (Firestore WebSockets): For real-time, ultra-fast synchronisation between offline distress payloads and the central dashboard.

3. Core Assumptions

For this solution to deploy flawlessly, these assumptions were established during architecture design:

  1. Low-Network Resiliency: We assume mobile network connectivity at crash sites is hostile. Therefore, we assume an offline-first service worker architecture is present on the client device, caching crashes and transmitting them via sub-50KB payloads immediately once any 1-bar EDGE/2G data signal is reestablished.
  2. Device Hardware Telemetry: The AI triage assumes the victim’s device (or IoT vehicle hardware) acts as an accelerometer and fetches basic physics data (G-force, Speed drops) alongside accurate GPS coordinates.
  3. Database Pre-Mapping (Global Applicability): We assume the integration of open global mapping (OSM APIs). This guarantees that police stations, ambulance rescue, towing services, and nearest puncture shops can be dynamically fetched in milliseconds, proving identical reliability whether the accident occurs in a rural Indian district or internationally.
  4. Data Protection (Rule 8.1): We assume strict anonymization of victim PII (Personally Identifiable Information). Crash data exported for future AI modeling (structured database) retains only physical telemetry, stripped of names or personal medical charts, complying with ethical tech tracking.

4. Code & Structured Database Delivery

  1. The full source code, including Python AI worker simulation (/data_pipeline), is packaged in our Github / ZIP submission.
  2. The structured database schemas required for the telemetry AI models have been submitted successfully as a raw NoSQL file configuration (database_models_schema.json) along with an export script, fulfilling Stage 1 submission criteria.