dissertation

CV_cut_identification is to try to identify the cuts on the suture mat using the realsense camera through cv code

yolov8 has my attempt at using yolov8 to identify wounds. create_dataset is to create a dataset of the suture pad under different lightings to be used for ML code

my_ML is a complete pipeline for training, validating, and testing an object detection model using PyTorch and the Faster R-CNN architecture on a COCO-formatted custom dataset.

Laparoscopic Suturing with UR3 in CoppeliaSim (ROS2 Humble)

This project integrates a UR3 robotic arm in CoppeliaSim with ROS2 nodes for laparoscopic suturing experiments. It combines simulation, computer vision, and inverse kinematics to reproduce suturing motions on a synthetic mold.

Repository Structure

ros_ws2/src ├── suture_arm │ ├── launch/ # ROS2 launch files │ ├── ML_detection/ # Machine learning models for cut detection │ ├── resource/ # STL models, scene files, and documentation │ ├── suture_arm/ # Main ROS2 package Python nodes │ │ ├── laparoscopic_ik_node.py # IK control for UR3 needle tip │ │ ├── vision_web.py # Vision + web interface node │ │ ├── stitching.py # Stitching logic │ │ └── (utilities and dataset scripts) │ ├── templates/ # Web UI templates │ └── test/ # Code style & testing └── ur_description # UR robot descriptions, meshes, and configs

Prerequisites

• Ubuntu 22.04 with ROS2 Humble installed

• CoppeliaSim (latest version)

• Python dependencies (install with pip):

  pip install numpy opencv-python torch torchvision

• Clone/download the ML detection models folder from: https://livemanchesterac-my.sharepoint.com/:f:/g/personal/jude_barakat_postgrad_manchester_ac_uk/EiiYwXlPeD1FgRe4GYUI2OoB5fPw_mqk-WBb6kzutxiBEQ?e=9peuTS

Place the downloaded contents inside: suture_arm/ML_detection/

Running the System

1. Open the simulation Launch ur3_suture.ttt in CoppeliaSim: ros_ws2/src/suture_arm/resource/ur3_suture.ttt

2. Build the workspace cd ~/Documents/Dissertation/Code/dissertation/ROS/attempt4/ros_ws2 colcon build

3. Source and run vision system (first terminal) source install/setup.bash ros2 run suture_arm vision_web

4. Source and run IK controller (second terminal) source install/setup.bash ros2 run suture_arm laparoscopic_ik_node

Components

• vision_web Publishes target stitching points from camera input via ML cut detection.

• laparoscopic_ik_node Drives the UR3 to align the needle tip with targets using damped-least-squares IK.

• stitching Handles stitching path generation and execution.

• ur_description Contains UR robot description files (meshes, configs, launch files).

Notes

• Ensure CoppeliaSim remote API is enabled (ZMQ API).

• If the robot collides or freezes, restart both CoppeliaSim and ROS nodes.

• For testing, you can run the demo launch:

  ros2 launch suture_arm suture_demo.launch.py

License

This project builds upon open-source packages. Check individual subfolders for license details.