OpenCV Pen Stealing Robot Arm


Overview

This project implements a vision-guided robotic system capable of autonomously writing and drawing with a pen. The system integrates perception, trajectory generation, and motion control within a ROS2 framework to convert visual input into precise end-effector motion.

The goal was to develop a complete perception-to-action pipeline for a structured manipulation task.


System Architecture

Perception

  • Camera-based detection of drawing workspace and target shapes
  • Image processing pipeline to extract drawing paths from input images or text

Planning

  • Conversion of image contours into ordered waypoints
  • Path smoothing and scaling to match workspace constraints
  • Pen up/down state management for continuous drawing

Control

  • ROS2 nodes for trajectory execution
  • Real-time communication between perception and motion modules
  • End-effector control to maintain consistent contact during drawing

Results

  • Successfully generated smooth trajectories for writing and shape drawing
  • Achieved accurate pen placement and consistent line quality
  • Demonstrated reliable perception-to-motion integration within a ROS2 system

The project showcases a full robotics pipeline combining computer vision, motion planning, and manipulation control.


Key Contributions

  • Developed contour extraction and path generation from images
  • Implemented ROS2-based modular architecture
  • Designed trajectory execution logic with pen state control
  • Integrated perception and control into a closed-loop workflow

Demo

Tracking demo


See also