Design and Implementation of an

Autonomous Sorting System Using

Mechatronics principles and AI
Contents
Objective .....................................................................................................................................2
Background .................................................................................................................................2
Problem Description ....................................................................................................................2
Deliverables.................................................................................................................................3
Evaluation Criteria .......................................................................................................................4
Additional Challenges ..................................................................................................................4
Simulation Software for Students .................................................................................................5
Quick List .................................................................................................................................5
Detailed Notes .........................................................................................................................5
Summary of Suggested Tools ....................................................................................................8
Cost details - Free or with discounted access ............................................................................8
Sensors details .......................................................................................................................... 10
Objective
The aim of this project is to design and develop an autonomous sorting system that efficiently
identifies, categorizes, and sorts objects of various shapes and sizes based on specific criteria
(e.g., size, color, or material type). This system should leverage mechatronics principles (such as
sensor integration, actuation, control systems, and mechanical design) and artificial intelligence
(AI) for real-time decision-making, system control, and process optimization.

The project should focus on the following mechatronics principles:

• Sensing and Measurement: Using vision or proximity sensors to detect objects.

• Actuation and Control: Designing mechanisms for sorting, such as robotic arms or
conveyor belts, that are controlled by feedback loops.

• Embedded Systems: Developing an embedded control system to handle inputs from

sensors and control the actuators in real-time.

• Automation and Robotics: Designing a system that can automatically handle different
objects without human intervention.

• AI and Machine Learning: Implementing AI algorithms to enable adaptive sorting based on

real-time data (e.g., object recognition or classification).

Background
In manufacturing, packaging, and recycling industries, automated systems are widely used for
sorting products, materials, or waste into categories for further processing. Traditional sorting
systems typically use pre-defined rules or manual intervention. With the advent of AI and
mechatronics, intelligent automated sorting systems are now becoming more flexible and adaptive,
enabling systems to handle dynamic environments, detect unknown objects, and optimize sorting
decisions in real-time.

This project will hands-on experience in integrating sensor technologies, mechanical actuators,
control systems, and AI to design and prototype a solution that can solve real-world industrial
automation challenges.

Problem Description
Develop an autonomous sorting system capable of identifying and sorting objects based on
specific attributes (size, color, shape, etc.). The system should consist of the following
components:

1. Object Detection: The system must utilize vision or proximity sensors (e.g., cameras,
infrared sensors, or ultrasonic sensors) to detect and recognize objects on a conveyor belt
or tray.
 2. Control System: Develop an embedded system (e.g., Arduino, Raspberry Pi) to interface
with sensors and actuators, making real-time decisions on sorting.

3. Sorting Mechanism: The mechanical part of the system should include actuators such as
robotic arms, servos, or pneumatic systems to perform the sorting action.

4. AI Integration: Use AI/ML models to classify objects based on their features (shape, size, or
color) and improve system performance over time by adapting to new objects.

5. Real-time Feedback and Adjustment: The system must be able to respond to dynamic
changes, such as moving objects with varying speeds, or objects that are not within the
original scope of the system.

The project must include:

• A mechanism to handle objects of varying weights and dimensions.

• A decision-making process that adjusts based on real-time data (e.g., AI model predictions
or sensor readings).

• A method of evaluating system efficiency, such as sorting accuracy or processing time per
object.

Deliverables
1. Project Documentation: A comprehensive report detailing:

o Design specifications (mechanical, electrical, and software architecture).

o Flow diagrams for control algorithms.

o Detailed AI model descriptions and training methods.

o Testing and validation results.

2. Prototype System: A working prototype demonstrating:

o Functional object detection and sorting.

o Integration of sensors and actuators.

o Real-time AI-based decision-making and feedback.

3. Presentation: A final presentation explaining the design process, challenges faced,

solutions implemented, and the impact of AI and mechatronics principles in the system.

4. Codebase: Complete and well-documented code for:

o Embedded systems controlling sensors and actuators.

 o AI algorithms for object classification.

Evaluation Criteria
1. Application of Mechatronics Principles (30%):

o Integration of sensors, actuators, and control systems.

o Effective use of embedded systems and automation techniques.

o Mechanical design of the sorting mechanism (e.g., stability, reliability).

2. AI Implementation (30%):

o Use of AI/ML algorithms for object classification.

o Real-time performance of AI models (speed and accuracy of sorting decisions).

o Adaptability of the AI to handle new object types or changing conditions.

3. Innovation and Creativity (20%):

o Creative solutions to real-world challenges such as object variability or dynamic

environments.

o Unique approaches to combining mechatronics and AI principles.

4. System Performance and Efficiency (20%):

o Sorting accuracy (e.g., percentage of objects correctly sorted).

o Processing time per object.

o Robustness of the system (e.g., system stability during extended operation).

Additional Challenges (to Expand Scope)

1. Multiple Object Types: Incorporate more than one sorting criterion (e.g., material type,
color, and size) and design a more complex AI model capable of handling multi-attribute
classification.

2. Multi-Robot Collaboration: Design a system where multiple robotic arms or sorting units
collaborate to handle a larger throughput, requiring coordination and communication.

3. Adaptability to Changing Conditions: Implement a feature where the system can adapt to
new types of objects or environmental changes (e.g., lighting conditions or object speed).

4. Energy Efficiency: Investigate methods to reduce power consumption by optimizing the

control algorithms or selecting low-energy actuators and sensors.
Simulation Software for Students (if physical hardware is not available):
Quick List
1. Tinkercad (Arduino Simulations): For simulating embedded control systems and basic
actuator/sensor interfaces.

2. Simulink (MATLAB): Ideal for modeling control systems, AI-based decision-making

algorithms, and mechanical systems (can simulate sensors, actuators, and control loops).

3. VEXcode VR: For simulating robots and sorting mechanisms in a virtual environment using
pre-built simulation tools.

4. Gazebo: A powerful robotics simulator for testing autonomous systems with 3D

environments, including vision-based sensors and robotic arms.

Detailed Notes
1. MATLAB/Simulink

• Description: MATLAB and Simulink provide powerful tools for modeling and simulating
mechanical systems, control systems, and AI applications.

• Mechatronics Simulation: You can use Simulink to model the robot arm, sensors, and
actuators, while MATLAB can be used for processing sensor data and controlling the sorting
actions.

• AI Integration: MATLAB offers toolboxes for machine learning, computer vision (with the
Image Processing Toolbox), and neural networks that can be used for color detection,
object classification, and process optimization.

• Pros: Great for algorithm development, real-time simulations, control systems, and AI
integration.

• Cons: Requires a license, which can be expensive.

Website: MATLAB/Simulink

2. V-REP (CoppeliaSim)

• Description: V-REP, now known as CoppeliaSim, is a powerful robotics simulation

platform that supports various types of robotic applications, including mechatronics
systems.

• Mechatronics Simulation: It allows you to simulate robotic arms, conveyors, actuators,

and sensors, as well as create real-time control loops.

• AI Integration: You can integrate Python scripts for AI algorithms, including computer vision
and machine learning, to recognize objects and make decisions based on color or shape.

• Pros: Highly versatile, supports multi-robot setups, and includes a built-in physics engine.
 • Cons: Learning curve for advanced functionality, but extensive documentation is available.

Website: CoppeliaSim

3. ROS (Robot Operating System) with Gazebo

• Description: ROS is an open-source robotics middleware, and Gazebo is a powerful

physics-based simulator often used in robotics research.

• Mechatronics Simulation: ROS can interface with sensors and actuators, while Gazebo
can simulate environments, robots, and real-world dynamics, allowing for testing of motion
control and sensor feedback loops.

• AI Integration: You can integrate AI algorithms, such as computer vision (OpenCV) and
machine learning (TensorFlow, PyTorch), for object recognition and classification, using ROS
nodes.

• Pros: Flexible, widely used in academic and research settings, large community, open-
source.

• Cons: Steep learning curve for beginners, requires setup and configuration of multiple tools
(ROS, Gazebo, etc.).

Website: ROS | Gazebo

4. Unity with ROS Integration

• Description: Unity is a widely used game engine that can be adapted for robotics
simulations. It is highly effective for visual simulations and allows for realistic rendering of
environments.

• Mechatronics Simulation: Unity can be used to model robots, sorting systems, and
environment layouts. You can integrate Unity with ROS for hardware simulation and control.

• AI Integration: Unity supports integration with AI tools, such as computer vision and
reinforcement learning, using TensorFlow or Python APIs.

• Pros: Highly realistic rendering, good for visualizing robotic operations, support for various
plugins, and easy to interface with ROS.

• Cons: Requires configuration for hardware interfaces and is not specifically built for
robotics, though with ROS integration, it becomes powerful.

Website: Unity

5. SolidWorks with SOLIDWORKS Simulation

• Description: SolidWorks is a professional 3D CAD software that can be used for designing
and simulating mechanical systems, including robotic arms, conveyors, and sorting
mechanisms.
 • Mechatronics Simulation: With SOLIDWORKS Motion, you can simulate the movement of
mechanical parts and analyze their dynamics. You can also simulate actuator forces and
kinematic behavior.

• AI Integration: While SolidWorks focuses on mechanical simulations, you can export

models to other platforms (e.g., MATLAB/Simulink or ROS) to integrate AI-based control
systems.

• Pros: Excellent for designing and simulating mechanical parts and systems, very precise
kinematic analysis.

• Cons: Not focused on AI or machine vision, so integration with other tools is necessary.

Website: SolidWorks

6. Pycharm with OpenCV and Simulation Libraries

• Description: If you prefer working in Python, you can use PyCharm (IDE) along with Python
libraries like OpenCV for object detection and PyBullet for physics simulation.

• Mechatronics Simulation: While OpenCV provides tools for color-based object detection,
PyBullet can simulate the robotic arm and environment interactions, allowing you to test
motion control and sorting mechanisms.

• AI Integration: OpenCV is widely used for computer vision tasks such as color detection
and classification, and you can integrate machine learning models using TensorFlow,
PyTorch, or scikit-learn for optimization.

• Pros: Full flexibility in programming and AI, open-source tools.

• Cons: You need to integrate different libraries and manage the environment setup yourself.

Website: PyCharm | OpenCV | PyBullet

7. SimScape (in MATLAB/Simulink)

• Description: SimScape extends Simulink to support multidomain simulation, including

electrical, mechanical, and hydraulic systems, making it useful for simulating mechatronics
systems.

• Mechatronics Simulation: It allows for modeling and simulating mechanical actuators,

sensors, and robotic systems, particularly useful for controlling physical systems.

• AI Integration: Use Simulink’s integration with AI and machine learning toolboxes to

implement optimization algorithms for sorting and task execution.

• Pros: Seamless integration with MATLAB, ideal for systems involving mixed-domain
simulation (e.g., mechanical and control systems).

• Cons: Requires MATLAB license.

Website: SimScape

Summary of Suggested Tools

1. MATLAB/Simulink: Ideal for control systems and integrating AI for sorting.

2. V-REP (CoppeliaSim): Good for robotic system simulation with flexibility.

3. ROS with Gazebo: Excellent for complex robotic simulations with AI integration.

4. Unity with ROS Integration: Best for visual simulations and integration with ROS.

5. SolidWorks: Excellent for mechanical design, with integration options for simulations.

6. PyCharm with OpenCV/PyBullet: Python-centric, good for deep integration with AI and
vision systems.

7. SimScape: Perfect for multidomain mechatronics systems simulation in MATLAB.

Each of these simulation tools can serve as a great platform to develop and test your automated
sorting system without the need for physical hardware, allowing you to focus on refining the
mechatronics and AI algorithms.

Cost details - Free or with discounted access

1. MATLAB/Simulink

• Cost: MATLAB/Simulink is not free; however, students can access it at a discounted price.

• Free Access: Many universities offer MATLAB/Simulink through campus-wide licenses

that allow students to access the software for free. Check with your institution to see if they
provide a free student license.

• Student License: If a campus-wide license is unavailable, students can purchase a

student license at a significant discount.

Website: MATLAB Student License

2. V-REP (CoppeliaSim)

• Cost: Free for students and academics.

• Free Access: CoppeliaSim is available for free for non-commercial use, which includes
student projects and educational purposes. However, commercial licenses require a fee.

Website: CoppeliaSim Free Version

3. ROS (Robot Operating System) with Gazebo

 • Cost: Free and open-source.

• Free Access: ROS and Gazebo are both open-source, which means they are completely
free to use for students and researchers. They have an extensive community and support
forums for help.

Website: ROS | Gazebo

4. Unity with ROS Integration

• Cost: Free for students and individuals with certain revenue limits.

• Free Access: Unity provides a free version for students and hobbyists, which can be used
for educational and non-commercial purposes. If you're developing a commercial product,
you would need to upgrade to a paid version.

• ROS Integration: You can integrate Unity with ROS using plugins, which are also free to use.

Website: Unity for Students

5. SolidWorks with SOLIDWORKS Simulation

• Cost: Not free, but discounted for students.

• Free Access: SolidWorks offers student versions at a discounted price, but it is not
completely free. Many educational institutions also provide access to SolidWorks through
campus licenses, so it's worth checking with your university for free access.

Website: SolidWorks Student Edition

6. PyCharm with OpenCV and Simulation Libraries

• Cost: Free for students.

• Free Access: PyCharm Community Edition is completely free for students. Additionally,
OpenCV and PyBullet are both open-source and free to use. You can freely access the
necessary libraries to work on simulations and AI applications.

Website: PyCharm Community Edition | OpenCV | PyBullet

7. SimScape (in MATLAB/Simulink)

• Cost: Not free, but available with student licenses.

• Free Access: Similar to MATLAB/Simulink, SimScape is available with student licenses at

a discounted rate. Some universities offer access through campus-wide licenses.
Website: SimScape Student License

Summary of Free Access Options for Students:

1. MATLAB/Simulink: Discounted student licenses and potentially free through campus

licenses.

2. CoppeliaSim (V-REP): Free for students for non-commercial use.

3. ROS with Gazebo: Completely free and open-source.

4. Unity with ROS Integration: Free for students, with certain revenue limitations for
commercial use.

5. SolidWorks: Discounted student licenses or campus licenses.

6. PyCharm with OpenCV/PyBullet: Free (open-source tools).

7. SimScape: Discounted student licenses and campus access.

Sensors details
These sensors can be used for object detection, classification, and automation in the sorting
system.

1. Vision-Based Sensors (Cameras and Image Sensors)

• Webcams (USB or Raspberry Pi camera):

o Use: For visual object detection and recognition (e.g., color, shape, and texture).

o Example: A USB webcam can be connected to a Raspberry Pi or computer to

capture images of objects on the conveyor belt.

o How It Works: The camera captures images of the objects, which can be processed
using image processing techniques (e.g., OpenCV) or AI models (e.g., convolutional
neural networks) for classification.

• Smartphone Cameras:

o Use: Smartphones are equipped with high-quality cameras and can be used for
image-based object detection.

o Example: A smartphone camera can be repurposed for detecting and classifying

objects based on color, size, or shape.

o How It Works: Using the phone’s processing power, image recognition algorithms
(e.g., TensorFlow or OpenCV) can identify and sort objects.
2. Proximity and Distance Sensors

• Ultrasonic Sensors (e.g., HC-SR04):

o Use: Measuring the distance of objects from the sensor.

o Example: These sensors are commonly used for proximity detection and can be
placed along the conveyor to detect the presence or distance of objects.

o How It Works: Ultrasonic waves are emitted, and the time it takes for the waves to
bounce back is measured to calculate distance, which is used for sorting decisions
or to prevent collisions.

• Infrared (IR) Sensors:

o Use: Detecting objects in close proximity (e.g., objects on a conveyor belt).

o Example: IR sensors can be used to detect objects based on infrared light

reflection.

o How It Works: When an object passes the IR sensor, the reflected light changes,
allowing the system to detect object presence or position.

• Time-of-Flight (ToF) Sensors:

o Use: To measure the distance to an object.

o Example: The VL53L0X ToF sensor can be used to measure the distance to objects
with high accuracy.

o How It Works: This sensor sends out a laser and measures the time it takes for the
laser to return, calculating the distance to an object.

3. Accelerometers and Gyroscopes (MEMS Sensors)

• Accelerometer (e.g., ADXL345):

o Use: Detecting motion, tilt, or orientation of objects or robotic arms.

o Example: An accelerometer can be attached to a robotic arm to ensure precise

sorting movements, detect tilt or misalignment, or monitor the belt's movement.

o How It Works: Measures the rate of acceleration along various axes, enabling
feedback for position or speed control.

• Gyroscope (e.g., MPU-6050):

o Use: Detecting rotational motion and angular velocity.

o Example: A gyroscope can help track and control the rotational movement of
actuators or robotic arms to ensure accurate positioning for sorting objects.
 o How It Works: It measures the rate of rotation around specific axes, helping to
stabilize or control motion.

4. Force and Pressure Sensors

• Force Sensing Resistor (FSR):

o Use: Detecting the force applied by an object (e.g., to identify when an object is
placed or pressed onto a sorting mechanism).

o Example: An FSR can be placed on a sorting tray to detect if an object has been
placed on it or when it has been moved.

o How It Works: As pressure is applied, the resistance of the sensor changes, which
can be measured to detect force or pressure levels.

• Piezoelectric Sensors:

o Use: Measuring pressure, vibration, or force.

o Example: Piezoelectric sensors can be used to detect vibrations or the impact of

objects in motion.

o How It Works: These sensors convert mechanical stress or vibration into an

electrical signal, which can be used to monitor objects during the sorting process.

5. Light and Color Sensors

• RGB Color Sensors (e.g., TCS3200):

o Use: Detecting color of objects to sort them based on color.

o Example: An RGB color sensor can be placed above the conveyor belt to detect the
color of objects and help classify them based on color for sorting.

o How It Works: The sensor uses light-sensitive elements to detect the intensity of
red, green, and blue light, which can then be used to classify the color of objects.

• Ambient Light Sensors (e.g., LDR):

o Use: Measuring ambient lighting levels to adjust the system's sensitivity in varying
light conditions.

o Example: A Light Dependent Resistor (LDR) can be used to adjust the sensor's
threshold for object detection based on environmental lighting changes.

o How It Works: LDR changes resistance depending on the amount of light it is

exposed to, which can be used to dynamically adjust system parameters.
6. RFID (Radio Frequency Identification)

• RFID Tags and Readers:

o Use: Identifying objects based on embedded tags, useful for advanced sorting
systems in warehouse automation or logistics.

o Example: Objects could be tagged with RFID labels, and an RFID reader would
identify the object before sorting.

o How It Works: The RFID reader sends out radio waves to communicate with the
RFID tags, retrieving stored information such as object type, size, or category for
sorting.

7. Motors and Actuators

• DC Motors / Stepper Motors:

o Use: Controlling the movement of sorting mechanisms such as conveyor belts,

robotic arms, or sorting trays.

o Example: A DC motor could be used to drive a conveyor belt, while a stepper motor
could be used to control the position of a robotic arm that picks and places objects.

o How It Works: DC motors provide continuous rotation, while stepper motors offer
precise control over position, making them ideal for precise object sorting tasks.

• Servos (e.g., SG90):

o Use: Precise control of small robotic arms or sorting mechanisms.

o Example: A small servo motor could be used to push or direct an object on the
conveyor belt to the correct sorting container.

o How It Works: Servos are controlled by sending a pulse-width modulated (PWM)

signal, which determines their position.

8. Environmental Sensors

• Temperature and Humidity Sensors (e.g., DHT11 or DHT22):

o Use: Monitoring environmental conditions in real-time, which could affect object

properties like weight or material consistency.

o Example: The system can adjust sorting parameters based on temperature and
humidity to optimize object handling.

o How It Works: These sensors provide real-time data on environmental conditions,

allowing adjustments to system behavior to ensure optimal performance.