Quality Control of Products on a Production Line using a 4 DOF Delta Robot

Project Description

The Quality Control of Products on a Production Line using a 4 DOF Delta Robot was a Senior Design Project - 2023 aimed at building an integrated inspection and sorting system that identifies and removes defective products from a production line using a Delta robot and machine learning.

The system leverages MobileNetV2 CNN for defect detection and a 4 DOF Delta Robot to remove defective products, minimizing human intervention and enhancing productivity. It combines real-time image processing, socket-based robotic control, and automated sorting in a streamlined pipeline.

Technologies Used

Python, OpenCV, TensorFlow/Keras (MobileNetV2), Custom Tkinter GUI, Socket Programming, 4-DOF Delta Kinematics, IR Sensors

Key Features

System Overview

Nakul Sharathkumar; Control System with Delta Robot and Conveyor
  • Integrated Workflow: Conveyor and IR sensors trigger real-time inspection and sorting operations.
  • MobileNetV2 CNN: Ensures reliable image-based defect detection with low computational load.
  • Autonomous Sorting: Delta Robot picks defective products upon IR sensor trigger and CNN result.

Delta Robot (4 DOF)

Nakul Sharathkumar; 4 DOF Delta Robot - Bottom View with Magnetic Actuator
  • Precision Picking: Magnetic end-effector safely removes faulty units from conveyor.
  • Repeatability: Mechanism accuracy demonstrated via grid test using laser pointer.
  • Socket Mirrored Simulation: Physical movements reflected in real-time on a Simulink model over a socket link.

Live Demonstration

  • System Demo: Full cycle of detection, prediction, and robotic action demonstrated.

My Contributions

As the Core Developer for this project, I designed and built the 4 DOF Delta Robot hardware and integrated a real-time socket interface to mirror its motion with a Simulink-based simulation model. I also developed a custom GUI using Tkinter to control and monitor robot behavior during testing and operation.

Project Outcomes & Impact

This project successfully demonstrated a functional prototype capable of classifying products using deep learning and autonomously handling defective items using robotics. The use of MobileNetV2 and Delta robot automation ensured accuracy, speed, and reliability.

It validated a real-time inspection and sorting system that can be adapted across various manufacturing domains for scalable, low-cost quality control solutions.

Team Acknowledgment

This project was a collaborative effort alongside a dedicated team, with each member contributing to the design, simulation, and overall development. Their expertise in mechanical design, control systems, and documentation played a vital role in bringing the prototype to life.

Team Members: