OpenArm Master–Slave Teleoperation

I engineered a low-cost, high-fidelity master–slave teleoperation system for robotic imitation learning, built on the OpenArm open-source platform. I custom-designed a master arm using Dynamixel actuators to capture precise human demonstrations, then extended the stack with VR teleoperation: 6‑DoF controller inputs are mapped to the slave over ROS 2 for intuitive data collection in diverse environments.

To turn this hardware into a complete autonomous pipeline, I log synchronized multi-modal data (RGB streams and joint states), train Diffusion / VLA policies with LeRobot, and run an end-to-end deployment path from teleop to on-robot policy execution. Isaac Sim complements the physical rig with VR teleop and dataset collection when iterating on pipelines or collecting data without the full hardware stack.

• Hardware: OpenArm-based slave; custom Dynamixel master; optional VR layer for demonstration capture.
• Software: ROS 2 integration; synchronized vision + proprioception logging.
• Learning: LeRobot; Diffusion Policy / VLA training; deployment to the slave arm.
• Simulation: Isaac Sim for VR teleoperation and data collection alongside real-world runs.

Media, figures, and repository links will be added as the project matures.