Autonomous Waste Sorting System (v2.0)

Virtual validation & digital twin for an autonomous, vision-based waste sorting line — Metal, Plastic, Glass, Paper, Battery, Organic.

---

Table of Contents

• Overview
• Problem & Solution
• Architecture & Technologies
• Technical Components
• Gallery
• Installation
• Usage
• Performance
• Project Structure
• Contributing & License

---

Overview

This project is an autonomous sorting system running on the Gazebo physics engine, designed to increase efficiency in waste management and reduce risks (e.g. battery leakage and fire) through virtual validation and a full digital twin of the physical process. Six waste categories are sorted without human intervention using computer vision and PID-controlled sweeper gates.

---

Problem & Solution

Aspect	Description
Industrial need	Manual sorting of lithium-ion batteries and mixed waste in recycling plants is slow, costly, and carries fire and toxicity risks.
Approach	Image processing and robotic control sort 6 waste types (Metal, Plastic, Glass, Paper, Battery, Organic) at high throughput and safely.
Digital twin	A complete software replica of the system allows design and tuning without hardware, and scales at low cost.

---

Architecture & Technologies

The stack uses a hybrid C++ / Python architecture: C++ for Gazebo plugins (physics, belt, spawners) and Python for ROS nodes (vision, control, dashboard).

Layer	Technology
Simulation	Gazebo 11
Middleware	ROS (Robot Operating System)
Vision	OpenCV & HSV-based color analysis
Control	PID controllers (P = 50.0) and Sweeper Gate mechanism
UI	PyQt5 real-time system dashboard

---

Technical Components

Node / Plugin	Role
/color_detector	Consumes camera images, runs HSV analysis, and publishes object class.
/sorter_controller	Computes gate trigger time Ttrigger from object speed and distance (millisecond accuracy).
conveyor_mover (C++ plugin)	Drives belt physics and object motion.
/system_dashboard	Live telemetry, statistics, and manual belt/gate control for operators.

---

Gallery

System Layout & Simulation Views

The simulation features a 9-meter conveyor belt with 6 distinct sorting stations.

Station 1 & 2	Full Line View	Conveyor Perspective
Sorting Logic	Side View	Top/Rear View

Computer Vision & Dashboard

Live Object Detection	Dashboard Interface

---

Installation

1. System requirements

• Ubuntu 20.04 LTS (or compatible)
• ROS Noetic or Melodic (desktop-full recommended)
• Gazebo 11

2. Workspace setup

mkdir -p ~/catkin_ws/src
cd ~/catkin_ws/src
git clone https://github.com/fuatsimsek/Autonomous-Waste-Sorting-System.git

3. Dependencies

sudo apt-get update
sudo apt-get install -y \
  ros-noetic-gazebo-ros-control \
  ros-noetic-effort-controllers \
  ros-noetic-joint-state-controller

4. Build & source

cd ~/catkin_ws
catkin_make
source devel/setup.bash

---

Usage

Launch the full stack (Gazebo world + ROS control + optional dashboard):

roslaunch pil_ayiklama_sim sorting.launch

Optional: run the gate test script in another terminal:

rosrun pil_ayiklama_sim test_gates.py

---

Performance

Simulation results at 0.6 m/s belt speed:

Metric	Result
Battery (project focus)	100% sorting success
Metal	100% sorting success
Overall detection	~96% success rate

---

Project Structure

Autonomous-Waste-Sorting-System/
├── assets/             # Screenshots: stand_1..6, camera_view_1..2
├── config/             # Controllers, RViz
├── launch/             # sorting.launch
├── scripts/            # Python nodes (detector, controller, dashboard, tests)
├── src/                # C++ Gazebo plugins (conveyor, spawner, etc.)
├── urdf/               # main_station.xacro
├── worlds/             # sorting_world.sdf
├── CMakeLists.txt      # Build configuration
├── package.xml         # Package dependencies
└── README.md           # Documentation

---

Contributing & License

Contributions are welcome. This project is under the MIT license.

---

Autonomous Waste Sorting System v2.0 — Digital twin & virtual validation on Gazebo + ROS.