🧠 Neural Networks | 📚 LLM/VLM | 🤖 Droids | 📱 Android

OpenAnimal is an extensible Unreal Engine–based framework for modeling and simulating animal- and human-like behavior. It integrates neural networks, behavior modeling, speech processing, and optional robotics interfaces directly within Unreal Engine.

OpenBrain is a self-contained, plug-and-play Animal Brain system designed to bring autonomous behavior to any Actor in Unreal Engine and to physical platforms. This module focuses on extensibility, aiming to simulate natural animal cognition and social interaction (rather than high-level intelligence).

The system is built to run fully offline with the brain on a phone and affordable hardware.

It is similar to robot operating systems like OpenBot or OM1 but has distinct features and a different philosophy.

• Native Unreal Engine ONNX neural-network inference in Blueprints
• Asynchronous processing pipelines
• Speech-to-Text and speaker recognition
• Animal pose, face, and gaze detection
  Estimate posture, orientation, and visual focus.
• Behavior and emotional-state modeling (preliminary)
  Early-stage behavior generation and emotion approximation.
• Platform-independent ONNX networks
  Single-file ONNX models that are openly available and operate consistently across platforms.
• Bottom-up autonomous intelligence modeling
  Designed to build emergent behavior from atomic perceptual and processing components.
• Android and Arduino based
  Enabling integration with broad variety of hardware setups

• Autonomous pets and droids
• Educational scenarios and demonstrations
• Robotics and hardware prototyping
• Research simulations, Social behavior studies and modeling

Latest release: v1.0.0

• ONNX Neural Networks - Now included in free plugin
  See: Neural Networks README for all single-file ONNX networks The networks are included now, but for convenience a Utility Widget is provided to automatically download the networks to the correct location.

• VaRest (will be replaced by ue in-house methods) Reliable HTTP/URL communication layer for interacting with hardware and large/visual language models.
  (VaRest is the battle-tested solution currently used, but it might be replaced by UE HTTP and JSON.)

• Ollama
  For local execution of large and visual language models, including support for cloud-based model computation via the new Ollama cloud-model options.

• Windows 64-bit (Editor, Runtime)
• Android (Runtime)
• (Linux Editor/Runtime)?

🚧 This project is a work in progress! It’s in an early-stage release, so some features are incomplete. Unfortunately, I can’t work on it full-time and need to focus on something more “lucrative”. I don’t have a donation system yet, but if you’d like to support this project—allowing me to dedicate more time and accelerate development—you can send me a note.

Email: OpenAnimal@outlook not .com but .de
For questions that may benefit others, please open an Issue here on GitHub