Week 7

in Blog

GSOC Coding Week 7 Progress Report

Week 7: Setting Up a Dual-Camera Simulation for Pick and Place

This week’s work centered on building the simulation pipeline for vision-based pick and place tasks using a UR5 robot and Robotiq 2F-85 gripper, integrated into the IFRA-Cranfield framework. The goal was to create a simulation environment that supports point cloud streaming and object detection for manipulation tasks.

What I Did This Week

  • Explored simulation packages:
    • Tested the IFRA_LinkAttacher plugin for simulating grasp and release actions
    • Built an initial pick-and-place task using the UR5 and Robotiq gripper

Package Exploration

  • Upgraded robot model with dual cameras:
    • Mounted one camera on the robot end-effector
    • Positioned a second camera fixed in the environment
    • Successfully published point cloud data from both cameras
  • Added objects to the simulation world:
    • Placed several target objects to prepare for vision-based manipulation

This setup forms the visual and physical context necessary for perception-guided motion planning.

Simulation Setup

Next Steps

For Week 8, I’ll focus on connecting the perception pipeline to the robotic action:

  • Run and debug the vision algorithm using live point cloud data
  • Localize objects for pick and place
  • Execute grasp planning and test robot execution flow

📍 Posted from Barcelona, Spain
🧠 Project: Migration and Enhancement of Machine Vision Exercise to ROS2 + MoveIt2