Technology Stack

Key Features

• Unreal Engine plugin with Blueprint-callable training and inference bridges
• Single and multi-environment support for reinforcement learning loops
• Python training module built on Gymnasium and Stable-Baselines3
• ONNX Runtime integration for running models inside the engine

About UE Reinforcement Learning

UE Reinforcement Learning

Why

I wanted a streamlined way to train and run reinforcement learning agents inside Unreal Engine while keeping the workflow accessible to both C++ developers and Blueprint users.

Interesting Technical Points

• Unreal Engine plugin with Blueprint-callable bridge classes for training and inference
• TCP handshake between Unreal and a Python module to exchange observation and action sizes
• Gymnasium wrappers and a Stable-Baselines3 training script with vectorized env support
• ONNX Runtime inference interface for running exported models locally

What I Learned

I learned how to design a clean UE plugin API, keep a Python training loop synchronized with an engine simulation, and ship an inference path that works in real time.