Week 15

in Blog

GSOC Coding Week 15 Progress Report

Week 15: Dropping the Planning Scene and Exploring Robotics Academy Structure

This week’s focus was on resolving planning scene issues and beginning the deeper integration into the Robotics Academy. After several attempts, I decided to temporarily move forward without the planning scene to avoid blocking progress.

What I Did This Week

  1. Attempted multiple approaches to fix the planning scene issue.
    • The problem remained that objects detected by perception were also included in the planning scene.
    • This caused false collisions and led to frequent manipulation failures during pick-and-place.
  2. Decision to remove the planning scene for now.
    • Since the scene was introducing more problems than benefits, I disabled it to ensure smooth execution of the exercise.
    • Pick-and-place runs more reliably without these false collision checks.
  3. Explored the Robotics Academy infrastructure.
    • Spent significant time studying how exercises are structured in the Academy.
    • Looked into the integration points between the exercise backend (ROS2 + MoveIt2) and the Academy’s GUI and APIs.
    • Gained familiarity with the directory structure, launch workflow, and Docker setup.

Problems I Encountered

  • The planning scene still shows dynamic objects, which causes unexpected collisions.
  • This blocked progress on manipulation tasks and slowed down development.

Solutions Implemented

  • Decided to move forward without the planning scene for now to maintain momentum.
  • Shifted focus to understanding the Robotics Academy structure for upcoming integration work.

Conclusion

img

Although the planning scene could not be resolved this week, removing it allowed pick-and-place manipulation to work again. More importantly, I now have a much clearer understanding of the Robotics Academy framework, which will be crucial for integration.

For Week 16, I plan to focus on:

  • Starting the integration of the exercise into Robotics Academy.
  • Connecting the backend pipeline (perception + manipulation) with the Academy interface.
  • Preparing the first prototype of the integrated exercise for testing.

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