Week 12

in Blog

GSOC Coding Week 12 Progress Report

Week 12: Gripper Fix and Integration with Perception Pipeline

This week’s work focused on fixing the gripper control and moving from hardcoded positions to perception-driven pick-and-place. These changes mark a big step toward making the workflow more autonomous and closer to the intended exercise functionality.

What I Did This Week

  1. Debugged and fixed the gripper control.
    • The issue was caused by a controller path misconfiguration, which prevented proper execution of gripper commands.
    • After correcting the setup, the gripper now operates reliably for attach/detach actions.
  2. Integrated the perception pipeline into the workflow.
    • Previously, pick-and-place sequences used hardcoded object and target positions.
    • Now, positions are obtained directly from the perception system, enabling dynamic detection of the object’s pose.
    • Verified that the robot can use these detected positions to successfully pick and place the object.
  3. Validated the updated workflow.
    • The pipeline now runs end-to-end: perception → motion planning → pick-and-place with gripper actions.
    • This removes the dependency on static positions and makes the exercise more realistic.

Problems I Encountered

  • Controller misconfiguration: Incorrect path mapping caused gripper failures.
  • Minor adjustments were needed to align detected positions with the motion planning pipeline.

Solutions Implemented

  • Fixed the controller path to ensure gripper commands execute properly.
  • Adapted the pipeline to consume perception outputs and feed them into the HAL API for motion execution.

Conclusion

Pick-and-Place Demo

With these updates:

  • The gripper works reliably during attach/detach.
  • The workflow now uses perception-driven object positions, replacing hardcoded values.
  • The system is closer to a fully autonomous pick-and-place exercise.

For Week 13, I plan to focus on:

  • Implementing the planning scene for better collision awareness.
  • Preparing the exercise for integration into the Robotics Academy infrastructure.

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