Robotics

T-shirt folding using robotic arm

Dual robotic arm system for automated T-shirt folding with ROS 2 and computer vision

IndustryRobotics
Year2024
Key technologiesROS 2, Python, ZED Camera
Headline resultSuccessfully folded T-shirts with 90% quality consistency matching manual folding
T-shirt folding using robotic arm

Overview

This project employs two robotic arms for T-shirt folding, aiming to pick and fold T-shirts from a pile. Controlled by ROS 2, the arms work together efficiently: one picks and positions the T-shirt, while the other folds it neatly. ZED cameras provide real-time visual data for detecting T-shirt sizes and colors, enhancing folding accuracy. Utilizing Python allows for versatility and customization to meet client needs. This innovative system boosts productivity and reduces manual labor, proving beneficial for apparel production and retail.

ROS 2PythonZED CameraComputer VisionRobotic Arms3D Perception

Key features

  • Dual robotic arm coordination with synchronized motion control
  • ROS 2 architecture for distributed system control and real-time communication
  • ZED stereo camera for 3D perception and depth sensing
  • Real-time T-shirt detection, size classification, and color recognition
  • Automated pick-and-place operations with adaptive gripper control
  • Intelligent folding sequence planning optimized for different garment types
  • Python-based control system for easy customization and client-specific requirements
  • Safety monitoring with collision detection and emergency stop capabilities

From challenge to solution

Challenge

Our solution

Coordinating two robotic arms to work in close proximity without collision

Implemented advanced trajectory planning with real-time collision avoidance algorithms

Handling fabric manipulation which is highly deformable and unpredictable

Developed specialized gripper designs with adaptive force control for gentle fabric handling

Detecting and classifying wrinkled T-shirts in various pile configurations

Created deep learning model trained on diverse T-shirt datasets for robust detection

Achieving consistent folding quality across different fabric types and sizes

Designed modular folding sequences that adapt to garment characteristics and feedback

Results & impact

  • Successfully folded T-shirts with 90% quality consistency matching manual folding
  • Reduced folding time by 70% compared to manual operations
  • Demonstrated adaptability across 15+ different T-shirt styles and sizes
  • Deployed in pilot retail operations with 95% uptime reliability
  • Achieved ROI within 18 months for medium-scale retail operations

Specifications

Arm Reach
850mm per arm
Payload
5kg per arm
Folding Speed
30-40 shirts/hour
Detection Accuracy
95% size/color
Camera Resolution
2.2MP stereo
Control Frequency
100Hz ROS 2

Gallery

Interested in similar work?

Tell us what you are building. We’ll help you identify the clearest next step.