Automatic Colour Sorting Machine

Interim Viva Report submitted for PROJECT-II (MECH 4295)

in the session 2024- 2025 (Even Semester) (4th Year)

By

Shri Chaitanya Agarwal

(Class Roll No.2157021)

Under the guidance of

Prof. Vishnukumar Singh

Assistant Professor

Department of Mechanical Engineering

Heritage Institute of Technology
 CONTENT
Topics Page No.
CHAPTER 1
1. Introduction 2

CHAPTER 2
2. Literature Review 5

CHAPTER 3
3. System Insights 6

CHAPTER 4
4. Components 9
CHAPTER 5
5. Conclusion 11

References 12

1
 CHAPTER 1

1. INTRODUCTION

Introduction and System Overview: Automated Lemon Sorting System Using AI-Based
Image Processing and Robotic Arm Integration

In the evolving landscape of precision agriculture, the need for efficient, accurate, and scalable
post-harvest processing systems has become more prominent. Manual sorting of fruits such as
lemons is labor-intensive, inconsistent, and limited in throughput. To overcome these issues, this
project presents an automated lemon sorting system that integrates mechanical design, AI-
powered image processing, and embedded control. The goal is to enhance productivity, ensure
uniform quality, and reduce human dependency in the agricultural supply chain.

The system is composed of several subsystems working in synchronization. It begins with a disk-
type rotary feeder powered by an MG996R 360-degree servo motor, which dispenses lemons one
by one from a hopper into the processing line. This motor’s motion is controlled by an Arduino
microcontroller, which ensures precise feeding of lemons onto a custom-designed slotted
conveyor belt.

The conveyor belt, tailored with lemon-sized slots, secures each fruit in place, maintaining
consistent orientation to support accurate imaging. As lemons progress on the conveyor, they
reach an IR sensor checkpoint. When a lemon is detected, the sensor sends a signal to the
Arduino to halt the conveyor. This controlled stop eliminates motion blur and ensures consistent
image capture.

At this stage, the lemon is positioned between two ESP32-CAM modules that capture high-
resolution side images simultaneously. These modules are compact, cost-effective, and
wirelessly transmit the images to a centralized AI-based image processing system. The dual-
camera setup improves classification accuracy by providing a more complete view of the
lemon’s surface.

The AI processing unit, trained using machine learning or deep learning techniques, analyzes the
images to classify lemons as ripe or unripe based on visual features like color and texture. Ripe
lemons typically appear bright yellow and smoother, while unripe ones have a greenish hue and
less uniform surfaces. The AI returns the classification result to the Arduino, which then
activates a robotic sorting mechanism.

A robotic arm with two degrees of freedom, powered by two MG996R 180-degree servo motors,
handles the physical sorting. Guided by the AI decision, the robotic arm picks the lemon from
the conveyor and places it in either the "Ripe" or "Unripe" bin. The entire action sequence is
executed in real-time and controlled by the Arduino.

2
This project showcases the integration of key engineering fields: mechanical design, embedded
systems, computer vision, and AI. Component choices, such as the ESP32-CAM and MG996R
servos, offer a balance between affordability and performance. The slotted conveyor enhances
the system’s reliability by ensuring proper alignment for imaging, which is critical for
classification accuracy.

A notable strength of the system is its modular and scalable design. Individual subsystems—such
as the vision module or sorting mechanism—can be upgraded independently. Higher-resolution
cameras, more sophisticated AI models, or additional conveyor lanes could be introduced to
increase capacity and accuracy without reworking the entire system.

From a software perspective, the Arduino operates as the control center, using a state-machine
approach to manage system flow through states like "Idle", "Feeding", "Detecting", "Imaging",
"Classifying", and "Sorting". This structured logic ensures each lemon is processed sequentially,
reducing errors and maintaining operational clarity.

Real-world applications for this system are extensive. In agriculture and food processing, where
quality control is crucial, automation can reduce labor costs, boost throughput, and ensure
consistent product standards. Additionally, as the AI system gathers more data, it can be
retrained to improve classification accuracy and even detect other fruit characteristics like
defects or size, enabling more granular sorting.

In summary, this automated lemon sorting system provides a practical and intelligent solution for
modern agriculture. It merges mechanical precision, real-time embedded control, and AI-based
image classification into a seamless workflow. With opportunities for adaptation to other
produce and future upgrades in sensing and AI, it represents a scalable, efficient, and forward-
thinking approach to post-harvest fruit sorting.

3
Process Block Diagram:

4
 CHAPTER 2

2. LITERATURE REVIEW
We have studied this following refrences and get targeted expertise about the work, and additionally talk
about some of them to analyze matters which required in this paper.
Ajay H. Chauhan [5] uses a sensor to sort lemons, addressing the growing demand for
automation in businesses. Sorting by color tones is challenging, and this study demonstrates fruit
selection using an Arduino NANO and TCS3200 color sensor. The design ensures high
precision, performance, and repeatability while being simple to construct, reducing costs, errors,
and labor.
Aye Myat Myat Myo [3] outlines a method for efficient color-based object sorting. The device
identifies RGB color balls and places them in appropriate bins. While some classifiers can't use
PIC microcontrollers, others leverage Arduino programming for improved efficiency. This
system reduces sorting time, minimizes human effort, and eliminates errors by those with color
blindness.
Ch. Shravani [2] utilized the TCS3200 color sensor with Arduino-controlled servomotors for
sorting. Sorting is labor-intensive and often inconsistent due to human fatigue and errors.
Machines excel at repetitive tasks, ensuring consistent performance. Unlike humans, who may
struggle with prolonged tasks or identifying colors accurately, machines maintain precision in
sequential processes.
N. Monika [4] introduces an automatic color sorting machine that sorts materials quickly and
accurately. Using a TCS3200 color sensor, the machine detects an object’s color and assigns it to
a specific box by controlling a servo motor. This system is useful in applications like grain
sorting in agriculture, food processing, mining, diamond industries, and recycling
Lim Jie Shen [1] conducted a background study as the first step, followed by classifying color
detection, assembling the robot body, and troubleshooting. The TCS3200D sensor detects colors,
the Arduino UNO controls operations, and the GS90 Tower Servomotor manages sorting via the
conveyor. Programming involved studying languages and comparing methods for successful
implementation.
Kunhimohammed C. K. [6] used two conveyor belts, a DC motor, an Arduino UNO, and a
TSC3200 sensor. One belt detects color, while the other sorts objects by color.
Rudresh H.G. [7] implemented a color-sorting robot with a robotic arm by a conveyor. The arm
sorts objects based on color, reducing manpower and increasing accuracy. A color sensor
identifies the color, sending signals to the microcontroller to operate the conveyor.
Shubham Gupta [8] describes a color recognition device comprising a light source, a sensor to
detect reflected light, a processing unit for color data analysis, and an output unit for results

5
CHAPTER 3

3. PROJECT INSIGHTS

1. Aim of the Project

 Develop a fully automated lemon sorting system.

 Classify lemons as ripe or unripe using AI-based image processing.
 Integrate mechanical, electronic, and software subsystems for smooth operation.
 Improve sorting accuracy, reduce manual labor, and increase throughput in post-harvest
processing.

2. Problem with Manual Sorting

 Manual sorting is labor-intensive, slow, and inconsistent.

 Variability in human judgment leads to quality issues.
 Not scalable for high-demand environments like farms or food industries.

System Architecture and Working Principle

3. Lemon Feeding System

 Lemons are loaded into a hopper at the beginning of the system.

 Hopper provides a buffered, gravity-assisted feed into the rotating feeder.

4. Rotary Feeder Mechanism

 A disk-type rotating feeder transports lemons to the conveyor belt.

 Driven by a 360-degree MG996R servo motor.
 Servo RPM controlled by Arduino for smooth, timed feeding.
 Ensures one lemon at a time moves to the conveyor.

5. Conveyor Belt System

 Lemons fall into a slotted conveyor belt, which holds them in place.
 Each lemon is securely positioned in a slot to prevent rolling/movement.
 Conveyor is Arduino-controlled to move lemons forward one by one.

6. Detection Using IR Sensor

 At a fixed point, an IR sensor is installed to detect lemons on the conveyor.

 As soon as a lemon reaches this point, the IR sensor sends a signal to Arduino.
 Conveyor immediately stops to allow accurate image capture.

7. Image Capture Mechanism

 Two ESP32-CAM modules are installed on both sides of the conveyor.

6
  They capture high-resolution images of the lemon's side views.
 Captured images are sent wirelessly to a computer or processing system.

8. AI-Based Image Processing

 A central computer performs image analysis using AI (ML/DL models).

 System is trained to distinguish ripe (yellow) and unripe (green) lemons.
 Features analyzed: color, surface texture, and uniformity.
 Output is a binary classification: Ripe or Unripe.
 The result is sent to the Arduino for final actuation.

9. Robotic Sorting System

 A robotic arm is installed at the end of the conveyor.

 Uses two 180-degree MG996R servo motors for actuation and flexibility.
 Controlled by Arduino based on AI output.

Sorting Logic:

 If Ripe → arm places lemon in Ripe Bin.

 If Unripe → arm places lemon in Unripe Bin.
 Entire pick-and-place process is automated.

10. Mechanical and Electrical Component Integration

 MG996R servos chosen for their high torque and reliable movement.
 ESP32-CAM used for its cost efficiency, compact design, and wireless transmission.
 Arduino used as a central controller for:
o Motor control (feeder, conveyor, robotic arm).
o Sensor monitoring (IR).
o Communication with AI processor.
o Decision-based actuation.

11. Software Architecture

 Entire flow is implemented as a finite state machine:

o Idle
o Feeding
o Detecting
o Imaging
o Classifying
o Sorting
 Ensures step-by-step handling of each lemon.
 No overlap or confusion in classification.

12. Advantages of Slotted Conveyor

 Ensures fixed spacing and orientation of lemons.

 Improves image capture accuracy.

7
  Reduces errors in robotic arm positioning.
 Provides mechanical consistency and repeatability.

13. Scalability and Modularity

 Each subsystem (feeder, conveyor, vision, sorting) is modular.

 Can be upgraded independently without redesigning the full system.
 Examples of upgrades:
o Higher-resolution cameras.
o More advanced AI models.
o Multiple robotic arms for increased speed.
o Parallel conveyor lanes for mass sorting.

14. Event-Driven Control System

 System responds to real-time events like:

o IR sensor detection.
o Image capture completion.
o AI classification result.
 Improves efficiency, accuracy, and system coordination.

15. Practical Applications

 Useful in agriculture, food processing, packaging, and retail.

 Can be adapted for other fruits like oranges, apples, tomatoes, etc.
 Reduces costs, increases productivity, and maintains uniform quality.

16. Long-Term Benefits

 AI model can continuously improve with more training data.

 Potential to classify lemons into more than two categories (e.g., overripe, damaged).
 Helps farmers and suppliers meet export standards and market requirements.
 Aligns with the trend of smart farming and Industry 4.0 automation.

17. Summary

 The project represents a smart, efficient, and accurate lemon sorting solution.
 Combines:
o Mechanical feeding
o Sensor feedback
o Image capture
o AI classification
o Automated sorting
 Designed to be scalable, low-cost, and easy to deploy.
 Demonstrates successful integration of mechatronics, AI, and automation in agriculture.

8
 CHAPTER 4

4. COMPONENTS:

 Hopper
Holds and guides bulk lemons into the rotating feeder in a controlled manner.
 Rotary Feeding Disk
Rotates to feed lemons one by one onto the conveyor belt using a timed mechanism.
 Slotted Conveyor Belt
Transports lemons forward while keeping them fixed in position and orientation using slots.
 ESP32 Camera
Captures high-resolution side images of lemons for AI-based ripeness analysis.
 IR Sensors. Detects when a lemon reaches the image capture zone (to stop conveyor).
 MG996R Servo Motor
Drives the rotary feeder (360° mode) and robotic arm joints (180° mode) for actuation &
Conveyor belt.
 Arduino
Central control unit managing motor movements, IR sensor inputs, camera communication,
and robotic arm control.
 Robotic Arm
Picks lemons from the conveyor and places them into ripe or unripe bins based on AI
classification.

9
Pending Tasks :

1. Hopper Manufacturing
o Status: Design completed
o Task: Fabricate or assemble the hopper based on the finalized design.
2. Rotary Feeder Manufacturing
o Status: Design completed
o Task: Build the rotary feeding disk with appropriate slots and mount it to the
servo shaft.
3. Conveyor Frame Construction
o Status: To be started
o Task: Design and fabricate the mechanical frame that holds and supports the
slotted conveyor belt.
o Notes: Ensure alignment with the rotary feeder and camera zones for consistent
lemon positioning.
4. Robotic Arm Support Link Installation
o Status: Link already bought
o Task: Mount the support link and assemble the robotic arm using the MG996R
servos.
o Notes: Calibrate for accurate reach and bin placement based on ripeness
classification.

10
 CHAPTER 5

5. CONCLUSION

The automated lemon sorting system is a smart, efficient, and cost-effective solution
designed to improve post-harvest processing in agriculture. By integrating mechanical
components with modern sensing and AI technologies, the system automates the entire
lemon classification process based on ripeness.

Lemons are first released from a hopper into a rotary feeder, which dispenses them one
by one onto a slotted conveyor belt. This ensures precise alignment and stability as
lemons move forward. IR sensors detect the presence of each lemon, pausing the
conveyor for inspection. Two ESP32 cameras capture side images, which are processed
using AI to determine whether the lemon is ripe or unripe based on color analysis.

The decision is sent to an Arduino, which controls the system’s actuators, including
MG996R servo motors and a robotic arm. The arm picks and places each lemon into the
correct bin, eliminating human intervention and improving accuracy.

With only a few tasks remaining—like hopper and rotary feeder manufacturing, conveyor
frame setup, and robotic arm mounting—the system is nearly ready for testing. This
project marks a significant step in agricultural automation, offering an intelligent and
scalable solution for quality-based fruit sorting.

11
REFERENCES

[1] Lim Jie Shen , Irda Hassan, “DESIGN AND DEVELOPMENT OF COLOUR SORTING
ROBOT”, Journal of Engineering Science and Technology

[2] Ch.Shravani, G. Indira, V. Appalaraju , “Arduino Based Color Sorting Machine using
TCS3200 Color Sensor” , Volume-8 Issue- 6S4 International Journal of Innovative Technology
and Exploring Engineering (IJITEE)

[3] Aye Myat Myat Myo , Zar Chi Soe, “Automatic Color Sorting Machine Using Arduino Mega
Microcontroller”, Volume VIII Issue VIII International Journal of Latest Technology in
Engineering, Management & Applied Science (IJLTEMAS)

[4] N. Monika, C.R. Pramod, B.S. Vinod, “Arduino Based Color Sorting Machine”, Volume 7
Issue 1 International Journal Of Innovative Research In Technology (IJIRT)

[5] Ajay H. Chauhan, Neema Ukani, Pratik Yennewar, Sandeep Sonaskar,Saurabh Chakole,
“Arduino Based Lemon Sorter Machine”, Volume 9 Issue 7 International Journal of Research in
Engineering and Science (IJRES)

[6] Kunhimohammed C. K, Muhammed Saifudeen K. K, Sahna S, Gokul M. S, “Automatic

Color Sorting Machine Using TCS230 Color Sensor And PIC Microcontroller”, Vol 2 Issue 2
International Journal of Research and Innovations in Science & Technology

[7] Rudresh.H.G, Prof. SHUBHA. “ Colour Sensor Based Object Sorting Robot”, Volume: 04
Issue: 08 International Research Journal of Engineering and Technology (IRJET)

[8] Shubham Gupta , Abhishek Rathee, Abhishek Kathariya, Harpreet Kaur Channi, “Modeling
and Designing of Color Detector using Arduino” , International Journal of Scientific Research in
Computer Science, Engineering and Information Technology

Arduino. (2015). Arduino Cookbook. O'Reilly Media.

Paul, R. (2019). Arduino Robotics: Building 10 robotic projects from the ground up. Packt
Publishing.

D’Ausilio, A. (2012). Arduino: A quick and easy guide to using Arduino for electronics projects.
CreateSpace Independent Publishing Platform

Adafruit. (n.d.). TCS34725 RGB color sensor with IR filter and white LED. Retrieved from
https://www.adafruit.com/product/1334

Hitec RCD. (n.d.). SG90 Micro Servo. Retrieved from https://hitecrcd.com/

12