Teach-and-Replay Implementation for Surface Erasing

Overview

This project was a group work at TUM-IMPL, focusing on the implementation of a teach-and-replay framework for the Franka FR3 robotic arm using :contentReference[oaicite:1]{index=1}.

The system enables kinesthetic teaching (manual guidance of the robot) and autonomous trajectory replay for a surface erasing task.

---

My Contributions

• Developed the main teach-and-replay Python script for:
  • Recording joint trajectories during manual guidance
  • Replaying trajectories using FollowJointTrajectory action
  • Supporting configurable replay speeds
  • CSV-based trajectory storage

• Integrated and configured a JointTrajectoryController for trajectory execution.

• Contributed to controller tuning, including PID gain adjustments to ensure stable trajectory tracking during the erasing task.

---

System Features

The implemented framework supports:

• Kinesthetic teaching via gravity compensation mode
• Autonomous replay through trajectory control
• Compatibility with:
  • Real FR3 hardware
  • Simulation (fake hardware mode)
• ROS 2–based modular structure

---

Technical Stack

• C++ and Python
• ROS 2 Control framework
• JointTrajectoryController
• Franka ROS 2 packages

---

This project strengthened my experience in ROS 2-based robotic system integration, trajectory control, and controller configuration for real-world manipulation tasks.