A comprehensive framework for reinforcement learning in robotics, which allows users to train their robots in both simulated and real-world environments.

A high-fidelity simulation backend for Franka robots that implements the complete libfranka API. Develop and test your robot control applications with identical code for both simulation and hardware. Supports all joint control modes (position, velocity, torque) with realistic physics powered by the Genesis engine.

MultiROS is an open-source ROS based simulation environment designed for concurrent deep reinforcement learning. It provides a flexible and scalable framework for training and evaluating reinforcement learning agents for complex robotic tasks.

RealROS is an open-source Python framework that seamlessly integrates with ROS (Robot Operating System) to create real-world robotics environments tailored for reinforcement learning (RL) applications. This modular framework simplifies RL development, enabling real-time training with physical robots

Goal-conditioned reinforcement learning like 🔥

Reinforcement Learning for Unmanned Airial Vehicles

The Tactile-MNIST Active Perception Benchmark

D4RL benchmark but ported to work end to end with gymnasium

Live Actor-Critic RL Training for CartPole

Training a MuJoCo Ant agent to walk using PPO from scratch.

Benchmarking of four continual-control RL algorithms - PPO, DDPG, TD3, and SAC - on gymnasium robotics environments.

Code for Edreate's Deep Reinforcement Learning Course: https://edreate.com/courses/deep-reinforcement-learning/

🎥 Record, replay, and analyze robot behavior across platforms with RoboReplay to improve robot execution and diagnostics.

Train a robot brain from zero to 100% success in 60 seconds flat — no hardware, pure NumPy, pure magic. Built on HuggingFace LeRobot standards.

Autonomous Rocket Landing with Deep Reinforcement Learning (Deep Q-Learning (DQN)) simulation in a custom Gymnasium environment inspired by SpaceX Falcon 9.

Documenting my journey to mastering Reinforcement Learning.

Record, replay, and diagnose robot behavior from any Python-based robotics stack.

A simulator with Gymnasium environments, for testing and developing multi-agent system algorithms.