RAVEN TELEOP

Gesture Based Tele-Operation Unleashed!
RAVEN TELEOP is versatile, low-cost, robot-agnostic, simulator-agnostic and requires no special hardware.

(CSCI5551 Course Project)

Demo

Full Demo Link: Demo

Introduction

RAVEN TELEOP is a middleware tele-operation framework for robots that translates 6-DoF hand gestures into real-time tele-operation commands, enabling seamless control across simulators like KinEval, PyBullet, Genesis, and Gazebo-ROS.

• Workflow:

Supported Platforms

Category	Supported Items
Simulators	✅ Gazebo · ✅ PyBullet · ✅ Kineval
Simulation Robots	✅ Turtlebot · ✅ Tugbot · ✅ Unitree A1 · ✅ MR2
Real-World Robots	✅ Turtlebot

Note: Moreover, you can follow codebase in /simulators to similary extend it to additional platforms.

Installation

1. Create the mamba environment (fast drop-in replacement for conda)

   mamba env create -f env.yaml
   mamba activate raven_env

Run

1. Server side: Run hand Landmark detection, create server & send data over ZMQ

   python landmark_detection.py
   

2. Client side: follow instructions in simulators/README.md to run our example simulated/real-world platforms.

License

MIT