Artificial Intelligence Based Waste Sorter

The world generates 2.01 billion tons of solid waste per year. This solid waste can be converted to energy. However, proper solid waste sorting is very important before converting it to energy. If waste is not sorted properly then toxic materials produce hazardous gases whi

Project Title

Artificial Intelligence Based Waste Sorter

Project Area of Specialization

Electrical/Electronic Engineering

Project Summary

The world generates 2.01 billion tons of solid waste per year. This solid waste can be converted to energy. However, proper solid waste sorting is very important before converting it to energy. If waste is not sorted properly then toxic materials produce hazardous gases which make the environment polluted. The objective of our project is to design and develop an Artificial Intelligence-based solid waste sorter. Our product will sort solid waste i-e, recyclable and non-recyclable materials. Firstly, we will generate a unique image dataset for objects present in solid waste. Including glass, paper, tin, and plastic. We will consider those objects only which are mostly recyclable material and are commonly present in domestic solid waste. Then, we will collect the images and divide them into two classes (recyclable and non-recyclable). After that, we will capture 500 to 600 images of each class. After collecting data from each glass, we will train the model on that unique dataset. Then we test and tune the model by using the Convolution Neural Network (CNN) model on the generated image dataset for classification. Then we put that trained model on raspberry pi 4. Our next step will be to design and manufacture (hardware) the sorter in the fabrication laboratory and install a Webcam to capture images in real-time. There will be a plate at the bottom of the hardware body at which material drop. That can rotate/slide clockwise and anticlockwise direction for dropping the material. There will be a Webcam that will be attached at the top of the plate and connected with a raspberry pi that captures the image of material on a real-time basis. The GPIO pins will turn on the Servo motor in a clockwise or anticlockwise direction to drop the material at their respective positions (bins) for recyclable and non-recyclable materials. In a nutshell, this project will improve the solid waste sorting for the Waste Management System.

In this project mainly we will use

• Raspberry pi 4
• Webcam
• Servo motor
• Python coding
• SOLIDWORKS software

In Pakistan, the government of Punjab has initiated the development of a waste-to-energy power plant of 35 MW capacity, fueled by municipal solid waste. Our AI-based solid waste sorter can be helpful for this waste-to-energy project in Pakistan. One of the major contributions of this project will be the generation of the dataset of recyclable objects in solid waste present in Pakistan. We will make this dataset freely available to the research community. The successful implementation of the project will reduce environmental pollution and contribute to the vision of green Pakistan.

Project Objectives

Aims

The aims of this project are:

• To know how different sensors and actuators including raspberry pi 4, camera, and servo motor work and how to get desired and optimum output from these devices.
• To understand how the deep learning convolutional neural network (CNN) classification model works and how to develop that model with a given dataset.
• To know about waste management systems and how we can make an effective waste management system to decrease the environmental pollution.
• To make an Artificial Intelligence (AI) based waste sorter to sort waste into two categories recyclable and non-recyclable waste.
• To make an Artificial Intelligence (AI) based waste sorter within given time and resources.

Objectives

Collection of solid waste data.

• Develop deep learning convolutional neural network (CNN) classification model on a given dataset.
• To design and fabricate the hardware body of the sorter.
• Interface camera with raspberry pi 4.
• Interface servo motor with GPIO pins of raspberry pi 4.

To synchronize the results of the trained model with the servo motor to get the desired output

Project Implementation Method

We have divided our project (Artificial Intelligence-based waste sorter) into three main stages. The stages of our project are;

• Collection of data and label that data.
• Construct a hardware body of sorter.
• Develop deep learning image classification CNN model to classify recyclable and non-recyclable waste.

 Collection of data:

• Build a dataset that contains almost 500-600 images per class.
• There are only two classes in our dataset recyclable waste and non-recyclable waste.
• The dataset will be improved with time to get better accuracy.        
• The whole results of our model depend on our data.                                                

Construct a hardware body of sorter:

• The structure of our hardware body will have two bins one for recyclable and the other for non-recyclable waste.
• There will be a metallic surface (a rectangular plate) the waste will fall on that surface and that surface will be connected to the servo motor. The surface will be able to move about 70 to 80 degrees in clockwise as well as anticlockwise directions.
• The surface will move in a clockwise direction it will dump waste in a recyclable bin or non-recyclable bin.
• At the top, we will have a camera the focus of our camera will be on the surface.
• For the surface need to be lit always so that camera can capture a clear image, we will put an LED bulb on top on the other side of the camera.

Develop deep learning image classification CNN model to classify recyclable and non-recyclable waste:

• Develop a deep learning image classification CNN model to classify waste into two categories recyclable waste and non-recyclable waste.
• Train the model on the given dataset.
• Download that model and place it in raspberry pi 4.
• Connect the raspberry pi with the camera and hardware body.

The GPIO pins are used to access the servo motor which will ultimately sort waste.

Benefits of the Project

Artificial Intelligence based waste sorter can play its role in waste management system and other related aspects including decrease global warming. With its roles it has some good benefits which are briefly described below.

• Reduce environment pollution:

The world generates 2.01 billion tons of solid waste per year. If waste is not sorted properly then toxics materials produces hazardous gases which make the environment polluted. Hence, using this project we will sort the recyclable and non recylable material which helps to reduces the environmental pollution as well as prevent the environment from hazardous gases.

• Helpful for waste to energy plants:

Our Artificial Intelligence based solid waste sorter will be helpful for this waste to energy project in Pakistan. The government of Punjab has initiated the development of a waste to energy power plant of 35 MW capacity, fueled by municipal solid waste. If we implement our project in those plants the performance and accuracy of the generation of energy will be increased and it also reduces labor cost.

• Reduces size of landfills:

Solid waste sorting helps to segregate the solid material into recyclable and non-recylable materials. The recyclable material would be used in different applications including waste to energy pants, recovered glass in asphalt to pave roads or recovered plastic in carpeting and park benches. Through this we can decrease the quantity of the solid waste materials and ultimately the size of landfills would be decreases.

• Promote recycling:

The recycled solid waste material can be used at many places as mentioned earlier waste to energy plants, recovered glass in asphalt to pave roads or recovered plastic in carpeting and park benches and prevent us from polluted and hazardous environment. This sorting play a vital role in development as well as sign of civilized nation. The new generation will seek the benefits of recycling and would dwell and inhale in pollution free environment.

• Generation of the dataset of recyclable objects:

We collect the unbiased dataset for solid waste to sort the recyclable as well non-recyclable solid waste. We will make this dataset freely available for research community. In future they can use that data to do more research on it and make something innovative which could be beneficial for the human as well the for the earth environment.

• Contribute to vision of green Pakistan:

There are some projects of government is ongoing. Including “Clean Green Pakistan” in which collection of solid waste, perform segregation and drop at their respective bins are main agenda. This is all about Solid Waste Management. By using this project we can contribute and lead this country to vision of green and safe Pakistan.

Technical Details of Final Deliverable

The aim of this project is to implement a fully functional prototype of the Waste Management System from the home level to the market level. The deliverable products would include the generation of datasets for recyclable materials and a prototype of an Artificial Intelligence-based waste Sorter which can be designed and printed in FABLAB Sukkur IBA University. This project is an integration of numerous technologies in a unified and robust framework. Artificial Intelligence-based waste sorting will provide

• Recyclable materials
• Non-recyclable materials
• Dataset of recyclable materials
• Real-time capturing using Webcam
• Offline working of the trained model on Raspberry pi 4
• Synchronization of Servo motor and Raspberry pi

And many other important deliverables.

Final Deliverable of the Project

HW/SW integrated system

Core Industry

Others

Other Industries

Energy , Health

Core Technology

Artificial Intelligence(AI)

Other Technologies

Clean Tech

Sustainable Development Goals

Good Health and Well-Being for People, Sustainable Cities and Communities, Climate Action

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com