TRIBHUVAN UNIVERSITY

INSTITUTE OF ENGINEERING
PASCHIMANCHAL CAMPUS

A Major Project Mid-Term Progress Report On

Delta Arm Waste Classifier

Submitted by:
Prateek Paudel (076/BEI/024)
Samman Shrestha (076/BEI/029)
Shiva Shrestha (076/BEI/038)
Sudarshan Gurung (076/BEI/044)

Submitted to:
DEPARTMENT OF ELECTRONICS AND COMPUTER ENGINEERING
PASCHIMANCHAL CAMPUS
LAMACHAUR, POKHARA

MANGSIR, 2080
 Acknowledgement
We would like to express our sincere gratitude to our dedicated project supervisor, Ms. Smita
Adhikari, for her unwavering support, valuable insights and continuous encouragement dur-
ing the mid-term stages of the Delta Arm Waste Classifier project. Her guidance has been
a beacon, explaining to us the intricacies of integrated technologies such as the YOLO v8
model. The clarity and depth of their feedback greatly contributed to the refinement of our
approach and the overall quality of our work. Her guidance continues to be the driving force
behind the success and development of our automated waste sorting and separation projects.

i
 ABSTRACT
The automated waste classification and sorting system presented in this project leverages
deep learning, specifically the YOLO (You Only Look Once) algorithm, combined with a
Delta Arm for efficient waste sorting. The system aims to streamline recycling processes
by automating the identification and separation of different waste materials. The core com-
ponents of the project include a camera module for image capture, a control unit for waste
classification using deep learning models trained on custom datasets, and a Delta Arm with
a vacuum suction pump for waste pickup and placement. The waste items are classified into
categories such as paper, plastic, glass, metal, and biodegradable waste. The YOLO algo-
rithm, implemented using Convolutional Neural Networks (CNN), enables real-time object
detection and classification. The coordinates of the waste items are calculated by the control
unit, and the Delta Arm is controlled by Arduino, guided by these coordinates. By combin-
ing the power of deep learning, object detection, and robotics, this system aims to enhance
waste recycling efficiency and contribute to sustainable waste management practices.

Keywords: Deep Learning, CNN, YOLO, Delta Arm, Object detection, Waste classification,
Waste sorting, Recycling

ii
TABLE OF CONTENTS

ACKNOWLEDGEMENT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i
ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii
LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi

1 INTRODUCTION 1
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Feasibility Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4.1 Technical Feasibility . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4.2 Economic Feasibility . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4.3 Operational Feasibility . . . . . . . . . . . . . . . . . . . . . . . . 3
1.5 Scope of Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.5.1 Hardware Development . . . . . . . . . . . . . . . . . . . . . . . . 4
1.5.2 Software Development . . . . . . . . . . . . . . . . . . . . . . . . 4
1.5.3 Training of Deep Learning Models . . . . . . . . . . . . . . . . . . 4
1.5.4 Integration of Robotics and AI . . . . . . . . . . . . . . . . . . . . 4
1.5.5 Performance Evaluation and Optimization . . . . . . . . . . . . . . 4

2 LITERATURE REVIEW 5

3 METHODOLOGY 6
3.1 Data Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.2 Model Training . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.3 Hardware Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.4 System Assembly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.5 System Block Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

4 SYSTEM DESIGN 8
4.1 Requirement Specification . . . . . . . . . . . . . . . . . . . . . . . . . . 8
4.1.1 Functional requirements . . . . . . . . . . . . . . . . . . . . . . . 8
4.1.2 Non-functional requirements . . . . . . . . . . . . . . . . . . . . . 8
4.2 Use Case Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

5 TOOLS AND TECHONOLOGIES 10

5.1 Hardware Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

iii
 5.1.1 Arduino Uno . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
5.1.2 Stepper motor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
5.1.3 Web CAM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
5.1.4 Air Suction Pump . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
5.2 Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
5.2.1 Python . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
5.2.2 TensorFlow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
5.2.3 OpenCV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
5.2.4 YOLO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

6 EPILOGUE 14
6.1 Expected Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
6.2 Budget Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
6.3 Work Schedule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

iv
List Of Figures

Figure 1: Block diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Figure 2: Use-case diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Figure 3: Arduino Uno . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Figure 4: Stepper Motor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Figure 5: Web CAM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Figure 6: Vacuum Air Suction Pump . . . . . . . . . . . . . . . . . . . . . . . . 12
Figure 7: Gantt Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

v
List Of Tables

Table 1: Budget Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

Table 2: Work Schedule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

vi
 CHAPTER 1: INTRODUCTION

1.1 Background

In today’s world, waste management and recycling have become critical issues for sustain-
able development. The increasing volume of waste generated by human activities calls for
efficient and effective waste sorting systems to minimize environmental impact and maxi-
mize resource recovery. According to the World Bank, global waste generation is expected
to increase by 70% by 2050 if current trends continue. Traditional waste sorting methods
often rely on manual labor, leading to time-consuming processes and inconsistencies in cat-
egorization. To address these challenges, advancements in robotics and artificial intelligence
(AI) offer promising solutions.
The ”Automated Waste Classification and Sorting” project aims to revolutionize waste man-
agement by developing an automated system capable of intelligently categorizing and sort-
ing different types of waste. By leveraging robotics and deep learning algorithms, the project
seeks to enhance recycling efficiency and contribute to a more sustainable future.
The Delta arm, renowned for its precision and speed, provides a versatile and agile robotic
solution for waste manipulation and placement. Through the utilization of a vacuum suction
pump, it enables accurate waste collection and transport, ensuring seamless handling of di-
verse waste items.
To identify and classify waste items, the project employs state-of-the-art deep learning al-
gorithms such as YOLO (You Only Look Once). By training a custom dataset, the system
can detect and recognize common waste categories, including paper, plastic, glass, metal,
and biodegradable waste. This classification capability enables the robot to make informed
decisions on the appropriate disposal location for each waste item.
The model processes the captured images from a camera module mounted on a conveyor
belt, calculates the waste item coordinates, and coordinates the actions of the Delta arm. The
integration of software frameworks and libraries, such as Python, OpenCV, and TensorFlow,
facilitates seamless communication and efficient coordination between the various compo-
nents of the system.
By combining robotics, deep learning, and precise waste categorization, the project aims
to create a reliable and efficient waste management solution. The ultimate goal is to sig-
nificantly reduce manual effort, streamline recycling processes, and promote a sustainable
approach to waste disposal.

1
1.2 Problem Statement

The efficient recycling of waste materials is crucial for environmental sustainability and re-
source conservation. However, in Nepal, the current percentage of waste that is effectively
being recycled remains significantly low. According to recent studies, only 4.1% of the total
waste generated in Nepal undergoes proper recycling processes. The major causes of low
recycling rate being improper separation of waste materials [1]. This indicates a significant
gap in waste management practices, leading to environmental pollution and the squandering
of valuable resources.
The existing waste management practices face significant challenges in achieving optimal re-
cycling rates and addressing the growing environmental concerns associated with improper
waste disposal. Traditional manual waste sorting methods suffer from inefficiencies, incon-
sistencies, and limited capacity to handle the increasing volume of waste generated world-
wide.
One of the key problems lies in the inadequate recycling of waste materials due to inefficient
sorting processes. Manual labor-intensive sorting methods often fail to achieve the required
level of precision and accuracy, resulting in cross-contamination and reduced recycling effi-
ciency.
The absence of automated technology hinders the ability to segregate paper, plastic, glass,
metal, and biodegradable waste effectively. As a result, recyclable materials are often dis-
carded in general waste streams, limiting their potential for reprocessing and reducing the
overall recycling rates.

1.3 Objectives

The main objectives of the ”Automated Waste Classification and Sorting” project is:
• To implement deep learning algorithms to train a custom model capable of detecting
and classifying different types of waste items, including paper, plastic, glass, metal, and
biodegradable waste.
• To build and utilize a Delta arm equipped with a vacuum suction pump to pick up waste
items and place them in their respective place.

2
1.4 Feasibility Analysis

The feasibility of our project can be evaluated in terms of technical, economical and opera-
tional feasibility.

1.4.1 Technical Feasibility

The utilization of well-established technologies such as robotics, deep learning algorithms,

and image processing techniques has been extensively researched and implemented in waste
management applications. For instance, research study conducted by Diya et al. (2018)
showcase the successful integration of robotics in waste sorting systems, highlighting the
technical feasibility of such solutions [2]. Additionally, the availability of open-source li-
braries and frameworks simplifies the implementation and integration of the required soft-
ware components.

1.4.2 Economic Feasibility

While there may be initial investment costs associated with hardware components like the
stepper motors, camera module, suction pump, control unit, the long-term cost benefits are
significant. By improving recycling efficiency, reducing manual labor requirements, and in-
creasing resource recovery rates, the project can lead to substantial cost savings in waste
disposal.

1.4.3 Operational Feasibility

One of the key advantages of the project lies in its potential to replace traditional manual
waste separation processes in companies. The operational feasibility of the project is evi-
dent in its ability to streamline and automate waste classification and sorting operations. By
employing robotics and AI-based algorithms, the system can accurately detect and classify
different types of waste items, eliminating the need for manual labor-intensive sorting.

1.5 Scope of Project

The scope of the ”Automated Waste Classification and Sorting” project encompasses the
development and implementation of a robust system for the automated classification and
sorting of waste materials. The project aims to create a sustainable solution that addresses
the challenges of manual waste separation methods and enhances recycling practices. The
key aspects of the project scope include:

3
 1.5.1 Hardware Development

The project involves the design and assembly of the necessary hardware components, such
as the Delta arm, camera module, suction pump, and control unit. The hardware components
will be integrated to form a cohesive system capable of efficiently picking up and sorting
waste items.

1.5.2 Software Development

The project includes the development of software components that enable waste classifica-
tion and sorting. This encompasses the implementation of deep learning algorithms, such as
YOLO, for accurate waste detection and categorization. Custom software modules will be
created to handle image processing, data analysis, and communication between the hardware
components.

1.5.3 Training of Deep Learning Models

The project involves the collection and preparation of a comprehensive dataset for training
the deep learning models. This dataset will include samples of various waste materials, such
as paper, plastic, glass, metal, and biodegradable waste. The deep learning models will be
trained using this dataset to accurately classify waste items.

1.5.4 Integration of Robotics and AI

The project aims to seamlessly integrate robotics and AI technologies to enable automated
waste classification and sorting. The robotic arm, controlled by the system, will pick up
waste items based on the classification results obtained from the AI algorithms. The system
will provide precise coordinates to guide the robotic arm in placing the waste items into their
respective containers.

1.5.5 Performance Evaluation and Optimization

The project includes a comprehensive evaluation of the system’s performance in terms of

waste classification accuracy, sorting speed, and overall efficiency. Iterative testing and opti-
mization will be conducted to enhance the system’s performance, fine-tune the deep learning
models, and ensure reliable operation in real-world scenarios.

4
 CHAPTER 2: LITERATURE REVIEW
The use of deep learning and delta arm for waste classification has been the subject of numer-
ous studies in the literature. A literature review for this project would involve researching
and analyzing existing studies, articles, and publications related to use of deep learning for
waste classification and delta arm.The review covers key aspects such as waste classification
methods, robotic systems in waste management, and AI algorithms for object detection and
classification.
Research by Zhang et al. explores the application of deep learning algorithms, including
YOLO, in waste detection and classification. Their work demonstrates the effectiveness of
such algorithms in accurately identifying different waste categories [3].
Research conducted by Satav et al. (2023) showcases the successful implementation of
robotic systems for waste sorting, emphasizing their potential to improve recycling rates
and reduce human labor requirements. The article highlighted the implantation of robotics
and AI in sorting of materials such as glass, paper, plastic, metals, etc., from other waste [4].
Robotic arms, including delta arms, have proven to be highly versatile and efficient in ma-
nipulating objects. They offer advantages such as speed, precision, and scalability, making
them suitable for waste handling and sorting tasks.
Another research conducted by Pierrot et al. (1990) highlights the use of delta arm in vari-
ous applications. Their work demonstrates the efficiency of delta arm in comparison to other
robotics arm. The main advantage of delta robots compared to traditional robotic arms is that
delta robots have motors attached to the main body instead of the arms. This allows delta’s
arms to move at high speed which makes the delta robot ideal for lightweight pick and place
operations [5].
Artificial intelligence, particularly deep learning algorithms, has revolutionized object de-
tection and classification tasks. Convolutional neural networks (CNNs), in particular, have
shown remarkable performance in accurately detecting and categorizing objects in images
and videos. Aishwarya et al. had shown the successfully implementation of The You Only
Look Once (YOLO) algorithm for waste classification, achieving high accuracy and effi-
ciency in identifying and categorizing different waste materials [6].

..

5
 CHAPTER 3: METHODOLOGY
The methodology for implementing the project ”Automated Waste Classification and Sort-
ing” involves several key steps, including data collection, model training, hardware integra-
tion, and deployment. These steps are essential to ensure the successful development and
implementation of an automated waste classification system.

3.1 Data Collection

The first step is to collect the necessary datasets for training the waste classification model.
This involves gathering existing waste classification datasets from sources such as Kaggle,
which provide labeled images of different waste materials. Additionally, custom datasets
will be created by capturing images of waste materials. It is important to ensure that the
datasets are diverse, containing samples representing various waste categories, lighting con-
ditions, and orientations. This will help the model generalize well and accurately classify
different types of waste.

3.2 Model Training

The next step is to train the waste classification model using the collected and labeled
datasets. The YOLO (You Only Look Once) model, known for its real-time object detec-
tion capabilities, will be utilized. The YOLO model architecture, such as YOLOv7, will
be implemented using deep learning frameworks like TensorFlow along with OpenCV. The
model will be trained on the prepared datasets, optimizing the model’s parameters through
iterations to improve its accuracy in identifying and classifying different waste materials.

3.3 Hardware Integration

Simultaneously, the hardware components required for the automated waste classification
system will be developed and integrated. This includes designing and building the Delta
arm, conveyor belt and connecting them together with an Arduino along with camera mod-
ule. The Delta arm will be responsible for picking up waste items based on the classification
results and coordinates obtained from the trained model. The camera module will capture
images of the waste items on the conveyor belt, providing input to the classification model.
The control unit will coordinate the operations of the hardware components and interface
with the software system.

6
3.4 System Assembly

In this phase, the software code and hardware components will be combined to create a cohe-
sive system. The trained waste classification model will be integrated with the control unit,
allowing for real-time waste classification. The system will process images captured by the
camera module, classify the waste items using the YOLO model, and provide precise coor-
dinates to guide the Delta arm in picking up and placing the waste items into their respective
containers. Extensive testing and fine-tuning will be conducted to ensure the system’s accu-
racy, reliability, and efficiency in waste classification and sorting.

3.5 System Block Diagram

Figure 1: Block diagram

7
 CHAPTER 4: SYSTEM DESIGN

4.1 Requirement Specification

The functional and non-functional requirements are:

4.1.1 Functional requirements

• Image Acquisition: The system should be able to acquire images of waste items from
the conveyor belt using the camera module.

• Object Detection and Classification: The system should accurately detect and clas-
sify waste items into predefined categories such as paper, plastic, glass, metal, and
biodegradable waste.

• Coordinate Calculation: The system should calculate the coordinates of each waste
item in the image to guide the Delta arm for picking and placing.

• Robotic Arm Control: The system should control the movements of the Delta arm,
enabling it to pick up waste items based on their coordinates and place them into their
respective containers.

• Real-time Operation: The system should process images and perform waste classifi-
cation in real-time to ensure efficient waste sorting.

4.1.2 Non-functional requirements

• Accuracy: The waste classification system should exhibit a high level of accuracy in
identifying and classifying waste items, minimizing misclassifications and errors.

• Speed and Efficiency: The system should perform waste classification and robotic arm
movements quickly and efficiently to achieve optimal throughput and productivity.

• Reliability and Robustness: The system should be reliable and robust, capable of han-
dling variations in waste items, lighting conditions, and environmental factors.

• Scalability: The system should be scalable, capable of handling a large volume of

waste items for classification and sorting.

• Maintainability: The system should be designed in a modular and maintainable man-

ner, allowing for future updates, enhancements, and repairs if needed.

• Safety: The system should adhere to safety standards to ensure the protection of users
and efficient operation in the waste sorting environment.

8
4.2 Use Case Diagram

Figure 2: Use-case diagram

9
 CHAPTER 5: TOOLS AND TECHONOLOGIES

5.1 Hardware Components

5.1.1 Arduino Uno

The Arduino Uno is an open-source microcontroller board based on the Microchip AT-
mega328P microcontroller (MCU). The Arduino Uno microcontroller board acts as the con-
trol unit for the hardware components in the project. It interfaces with the PC and receives
instructions for controlling the stepper motor and conveyor belt. The Arduino Uno executes
these commands to move the Delta arm and regulate the movement of the conveyor belt.

Figure 3: Arduino Uno

5.1.2 Stepper motor

A stepper motor is a type of brushless DC electric motor that converts digital pulses into
precise mechanical movements. It is called a ”stepper” motor because it moves in discrete
steps or increments, allowing for precise control over its position and rotation. It receives
commands from the Arduino Uno to rotate and position the arm accurately. The stepper
motor’s precise steps allow for controlled and repeatable motion, enabling the Delta arm to
pick up waste items with precision.

10
 Figure 4: Stepper Motor

5.1.3 Web CAM

A webcam is an input device that captures digital images. A webcam or camera module
is employed to capture images of the waste items on the conveyor belt. The webcam is
connected to the PC and captures real-time video feed, which is then processed for waste
classification and coordinate calculation.

Figure 5: Web CAM

11
 5.1.4 Air Suction Pump

An air suction pump is integrated into the Delta arm to create a vacuum suction mechanism.
This suction power facilitates the firm grip and lifting of waste items. It consists of a DC
motor to generate vacuum and a solenoid valve with a suction cup.

Figure 6: Vacuum Air Suction Pump

5.2 Software

5.2.1 Python

Python is a popular programming language used for developing various applications, includ-
ing machine learning and computer vision. It serves as the primary programming language
for implementing the waste classification and sorting system. Python provides a wide range
of libraries and frameworks that facilitate image processing, deep learning, and communica-
tion with hardware components.

5.2.2 TensorFlow

TensorFlow is an open-source machine learning framework widely used for building and
training deep learning models. It provides a comprehensive set of tools and functionalities
for creating and deploying neural networks. In this project, TensorFlow is utilized for train-
ing and implementing the YOLO (You Only Look Once) model.

12
 5.2.3 OpenCV

OpenCV is a library that provides various computer vision functions and algorithms. It
offers a wide range of tools for image processing, feature detection, and object tracking. In
this project, OpenCV is used for calculating the coordinates of bounding box form predicted
image.
5.2.4 YOLO

YOLO is a Convolutional Neural Network (CNN) for doing object detection in real-time.
The algorithm applies a single neural network to the full image, and then divides the image
into regions and predicts bounding boxes and probabilities for each region. It is widely used
for applications such as object recognition, autonomous driving, and surveillance systems. In
this project, YOLO is employed for waste classification, allowing the system to identify and
categorize different types of waste items, such as paper, plastic, glass, metal, and biodegrad-
able waste. Different versions of YOLO are availabe. We will be using YOLO v8 which is
latest version for our project.

13
 CHAPTER 6: EPILOGUE

6.1 Expected Output

The expected output of the Automated Waste Classification and Sorting system includes ac-
curate waste classification results for each item detected on the conveyor belt, providing the
identified waste category (paper, plastic, glass, metal, or biodegradable waste). The system
also calculates and provides precise coordinates for each waste item, guiding the Delta arm
in picking up and placing them into their respective containers. The system controls the
robotic arm movements in real-time, ensuring efficient and accurate waste sorting.

6.2 Budget Analysis

Component Quantity Total Price(Rs.)

Arduino Uno 1 1000
Stepper-Motor 3 6000
DC Worm Geared Motor 1 1000
Stepper-Motor Driver 3 900
Motor Driver 1 500
Vacuum Suction 1 3000
Conveyor Belt 1 1000
Camera 1 1200
Frame Components - 5000
Miscellaneous - 2500
Total 22100

Table 1: Budget Estimation

14
6.3 Work Schedule

Figure 7: Gantt Chart

Task Start Date Duration (Days)

Project Initiation 6/7/2023 7
Dataset Collection 6/15/2023 15
Model Training 7/1/2023 20
Hardware Integration 7/1/2023 60
System Assembly 10/15/2023 75
Testing 1/1/2024 30
Deployment 2/1/2024 10
Documentation 6/10/2023 280
Maintainance 2/11/2024 15

Table 2: Work Schedule

This means that the entire project should take approximately 9 months to complete. How-
ever, the exact duration will depend on the complexity of the system and the availability of
resources.

15
 REFERENCES
[1] A. Khatoon, “Waste management—a case study in nepal,” Solid Waste Policies and
Strategies: Issues, Challenges and Case Studies, pp. 185–196, 2020.

[2] S. Z. Diya, R. A. Proma, M. N. Islam, T. T. Anannya, A. Al Mamun, R. Arefeen,

S. Al Mamun, I. I. Rahman, and M. F. Rabbi, “Developing an intelligent waste sort-
ing system with robotic arm: A step towards green environment,” in 2018 International
Conference on Innovation in Engineering and Technology (ICIET). IEEE, 2018, pp.
1–6.

[3] Q. Zhang, X. Zhang, X. Mu, Z. Wang, R. Tian, X. Wang, and X. Liu, “Recyclable waste
image recognition based on deep learning,” Resources, Conservation and Recycling, vol.
171, p. 105636, 2021.

[4] A. G. Satav, S. Kubade, C. Amrutkar, G. Arya, and A. Pawar, “A state-of-the-art review

on robotics in waste sorting: scope and challenges,” International Journal on Interactive
Design and Manufacturing (IJIDeM), pp. 1–18, 2023.

[5] F. Pierrot, C. Reynaud, and A. Fournier, “Delta: a simple and efficient parallel robot,”
Robotica, vol. 8, no. 2, pp. 105–109, 1990.

[6] A. Aishwarya, P. Wadhwa, O. Owais, and V. Vashisht, “A waste management technique

to detect and separate non-biodegradable waste using machine learning and yolo algo-
rithm,” in 2021 11th International Conference on Cloud Computing, Data Science &
Engineering (Confluence). IEEE, 2021, pp. 443–447.

16