danielpmorton

danielpmorton / frax

Public

Fast Robot Kinematics and Dynamics in JAX

45
0
100% credibility
Found Apr 08, 2026 at 46 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

frax is a Python library that quickly computes robot positions, speeds, forces, and collision risks for building responsive robot controllers.

How It Works

1
🔍 Discover frax

You hear about frax while looking for a simple way to make robots move smoothly and calculate their forces.

2
📥 Set it up

Download and install it on your computer with one easy command, no complicated setup needed.

3
🤖 Pick your robot

Choose a ready-made robot like a helpful arm or a walking humanoid to start playing with.

4
📐 Describe the robot

Share a simple file telling it your robot's shape, joints, and weights so it understands perfectly.

5
⚙️ Test movements

Ask it to figure out things like how the robot balances or reaches for objects.

6
🚀 Build a controller

Follow fun examples to create smooth motions that avoid crashes and feel natural.

🎉 Super-fast robot ready

Your robot now computes everything lightning-quick, perfect for real-world control!

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 46 to 45 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is frax?

frax loads URDF robot models into Python using JAX for ultra-fast kinematics and dynamics: mass matrices, Jacobians, gravity, Coriolis terms, even collision distances via sphere approximations. Expect low-microsecond CPU times for 25-100kHz control loops or 100M+ computations/sec on GPU/TPU, with seamless JIT and autodiff for gradient-based controllers. Preloads Franka Panda for fast robot arms and Unitree G1 for fast robot dogs.

Why is it gaining traction?

Matches Pinocchio speeds with Python simplicity and JAX parallelism/differentiation—no more manual derivations for IK, OSC, or CBFs enforcing singularity/joint-limit/collision avoidance. Demos track sinusoids or dodge obstacles interactively, scaling to batches for sim-to-real. Stands out for real-time prototyping without C++ builds.

Who should use this?

Robotics engineers building MPC/RL controllers for fast robot arms or humanoids, especially in JAX/MuJoCo stacks needing diff dynamics. Suited for researchers optimizing safe trajectories on Unitree G1s or Panda manipulators, or anyone ditching Pinocchio's overhead for GPU-accelerated fast GitHub runners.

Verdict

Grab it for a fast GitHub download if you're in JAX robotics—examples and docs shine despite beta flux. 45 stars and 1.0% credibility signal early days; test thoroughly before production.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.