🤖 net_franky

Use the Franky library from non-realtime Machines.

🏗️ How it works

┌─────────────┐    RPyC     ┌─────────────┐    libfranka    ┌─────────────┐
│ Your Laptop │ ──────────► │ RT Server   │ ──────────────► │ Franka Robot│
│             │             │             │                 │             │
└─────────────┘             └─────────────┘                 └─────────────┘

🎯 Key Benefits:

• 🔌 Drop-in replacement for the franky library with function stubs
• ⚡ Well proven - building on the common rpyc library
• 🚀 Simple - 5 minutes to get started

🚀 Quick Start

1. Server Setup (Real-time machine)

# One-time setup
mkdir ~/net_franky && cd ~/net_franky
python -m venv .venv && source .venv/bin/activate
pip install "net_franky[server] @ git+https://github.com/yblei/net_franky.git"

2. Start server (Run in tmux/screen)

rpyc_classic -p 18812 --host 0.0.0.0

3. Client Usage (Your laptop)

pip install 'git+https://github.com/yblei/net_franky.git'

from net_franky import setup_net_franky
# Connect to remote server
setup_net_franky("server-ip", 18812)

from net_franky.franky import Robot

# Use exactly like local franky
robot = Robot("192.168.1.100")
robot.move_to([0.3, 0.4, 0.5])

That's it! 🎉 Your robot code now runs remotely.

Only Deviation from franky: Wrapper around the robot class

Franky supports the registration of a callback function to a motion. This is useful to record trajectories or to stream pose information back for visualization. Since this function is called with 1000Hz and the calls are buffered, execution over the network leads to significant delays.

We therefore provide a pointer to the latest motion callback data in

robot_state, time_step, rel_time, abs_time, control_signal = robot.get_last_callback_data()

Warning: If you get segmentation faults, make sure you use the same version of python on the server and the client.

🤝 Contributing

We welcome contributions!