Skip to main content

The research layer of SuperDex

Welcome to SuperDex Lab.

Early Preview

SuperDex Lab connects simulation and policy development through a Gymnasium-style API for reinforcement learning. It is currently in early preview and will receive substantial improvements. A general abstraction for partially observable Markov decision processes will underpin applications in reinforcement learning, system identification, and model predictive control.


A shape-sorting manipulation policy trained in simulation with SuperDex Gym and deployed directly on a real-world robotic hand.

SuperDex Gym

A Gymnasium-compatible reinforcement-learning framework built on the SuperDex Physics engine. Train agent controllers in high-fidelity simulation with a suite of classic-control and locomotion environments.

RLlib Training

An off-the-shelf Ray/RLlib integration for RL training straight from the Gym environments.

Get started with SuperDex Lab →