CoolFace
Apppublic

C-S-F/CyberCollect

sourceHugging Faceupdated 7mo agoView on Hugging Face
0likes
App README

<h1 align="center"> <br> <img src="https://img.icons8.com/nolan/256/satellite.png" alt="CyberCollect" width="120"> <br> CyberCollect 🛰️ </h1>

<h4 align="center">A Production-Grade Cyberpunk 3D Route Optimization & Waste Management Platform</h4>

<p align="center"> <a href="#key-features">Features</a> • <a href="#architecture">Architecture</a> • <a href="#the-5-levels-of-optimization">The 5 Levels</a> • <a href="#quick-start">Quick Start</a> • <a href="#deployment">Deployment</a> </p>

🌌 Overview

CyberCollect transforms a multi-tiered waste collection optimization project into a stunning, interactive, cyberpunk-themed Web application.

Built with FastAPI on the backend and native Three.js + HTML/JS/CSS on the frontend, the application visualizes complex algorithmic data—ranging from Dijkstra graphing to VRP Tabu searches—on a glowing 3D WebGL satellite map. It also features an integrated Local LLM AI Assistant to analyze data and operations in real-time.

CyberCollect Concept FastAPI Three.js Ollama

🚀 Key Features

  • —Cyberpunk 3D Visualization: Natively rendered Three.js scene mimicking a high-tech tracking satellite. Includes neon grids, bloom effects, animated routes, and pulsing alert nodes.
  • —Multi-Level Algorithmic APIs: 5 distinct Python optimization engines exposed via RESTful APIs.
  • —Live Dashboard & Visualizers: Real-time generation of Gantt charts, radial utilization bars, and distance convergence curves synced to 3D map animations.
  • —Local AI Assistance (Ollama): Slide-in chat panel with Server-Sent Events (SSE) streaming. The AI is automatically injected with the contextual state of whatever algorithm is currently running on-screen.
  • —Zero-Cost Native Deployment: Configured via Docker & Render Blueprints to run completely free ($0/mo) in the cloud.

🧠 The 5 Levels of Optimization

The backend handles five distinct, progressively complex operational tiers of municipal waste logic:

  1. 1.Level 1 (Shortest Path): Constructs the road mesh using Dijkstra's algorithm to analyze raw travel distance and constraints between collection nodes.
  2. 2.Level 2 (Greedy Assignments & Load Balancing): Triages incoming zone demands and optimally assigns them to truck pools using greedy capacities and variance minimization.
  3. 3.Level 3 (Tri-partite Schedule Planning): Solves time-window constraints (traffic congestion, legal driving hours, night restrictions) and generates full weekly truck collection schedules.
  4. 4.Level 4 (Vehicle Routing Problem - VRP): Uses 2-Opt & Tabu Search heuristics to generate exact stop-by-stop sequencing and turn-by-turn routing for each truck in the fleet.
  5. 5.Level 5 (Real-time IoT Simulation): A live streaming engine ticking by the minute. Simulated IoT bin sensors trigger ALERTE_REMPLISSAGE events, forcing live tactical dashboard updates.

🏗 Architecture

mermaid
graph TD
    A["Browser: Cyberpunk SPA"] -->|REST/SSE| B["FastAPI Backend"]
    B -->|Python Backend| C["L1: Dijkstra"]
    B -->|Python Backend| D["L2: Truck Assignment"]
    B -->|Python Backend| E["L3: Weekly Planner"]
    B -->|Python Backend| F["L4: VRP Opt"]
    B -->|Python Backend| G["L5: IoT Simulation"]
    B -->|HTTP SSE| H["Ollama LLM (local)"]
    B -->|Serves static files| A

    subgraph "Frontend (Three.js)"
    A --> I["3D Satellite Map"]
    A --> J["Algorithm visualizers L1-L5"]
    A --> K["AI Chat Drawer"]
    end

📦 Quick Start (Local Run)

You can run CyberCollect directly on your local machine using Python or Docker.

Option A: Python / Uvicorn (Recommended)

  1. 1.Clone the repo:
bash
   git clone https://github.com/desagencydes-rgb/CYBERCOLLECT.git
   cd CYBERCOLLECT
  1. 1.Install requirements:
bash
   pip install fastapi uvicorn httpx python-multipart
  1. 1.Run the server:
bash
   cd projet_collecte_dechets
   uvicorn webapp.backend.main:app --host 0.0.0.0 --port 8000 --reload
  1. 1.Open Application at http://localhost:8000

Option B: Docker Compose

bash
docker-compose up --build

🤖 Activating the AI Assistant

The AI chat is an optional progressive enhancement. The app dynamically pings for an LLM on startup. If found, the chat interface is unlocked.

  1. 1.Download Ollama for your OS.
  2. 2.Pull a lightweight model in your terminal:
bash
   ollama pull llama3.2:3b
  1. 1.Ensure the Ollama daemon is running (ollama serve).
  2. 2.Hard Refresh the CyberCollect Web App. The AI Badge will flip to Green and the system will greet you!

☁️ Free Deployment ($0)

This project has been explicitly engineered for $0 free-tier cloud deployment using Render.com.

  1. 1.Fork this repository.
  2. 2.Go to Render Dashboard.
  3. 3.Click New + > Web Service.
  4. 4.Connect your GitHub fork.
  5. 5.Render will automatically detect the render.yaml Blueprint file and deploy both the FastAPI Backend and the Web UI in seconds.
Note: Render's free tier instances do not possess the GPU/Memory to run the Ollama LLM docker image. Consequently, the cloud-deployed version of CyberCollect will auto-detect the lack of an LLM and display a graceful local-fallback message in the AI Drawer.

<p align="center"> <i>"Efficiency is just data organized correctly."</i><br> <b>— CyberCollect OS</b> </p>