Projects Made Using HARFANG® Tools


An air-to-air combat sandbox simulation, created in Python 3 using HARFANG® Framework, featuring advanced game mechanics and environments such as : custom ocean, terrain & skydome shaders, dynamic clouds, an autopilot mode (take-off, landing, taxi), and a network mode allowing users to control the aircrafts from a third-party machine.
The project leverages the interoperability of the Python-based air simulation sandbox to develop reinforcement learning for aircraft piloting. The Sandbox has served as a basis for several academic publications.

Poppy Ergo Jr Digital Twin

This project demonstrates an application of the HARFANG®3D API in Python with the Poppy Ergo Junior robot. The goal was to create an accurate digital twin of the robot, that users could easily interact with. Two modes allow the articulated Poppy to be moved around, either by interacting with the 3D model with touch control via an external monitor, or by handling the actual physical robot.

SNCF Human Factor Study in VR

This project aims to recreate of a train station in virtual reality to extract behavioral data from a panel of users. The combination of a fully equipped VR facility and a Python VR platform based on HARFANG allows to study the effects of equipments or methods on the behavior of users. The analysis of the collected data helps to measure and quantify to what extent the experiential situation has an effect on the perception, understanding and behaviour of the user:
- Eye-tracking data, represented as a heatmap point cloud
- Data on the kinematics of the movements
- Physiological data, heart frequency, pupillary dilatation
- Audiovisual recordings, multi-view cameras, real and VR

Reachy Digital Twin

Digital twin of the Reachy robot. Reachy is designed an built by the French company Pollen Robotics as an open-source robotic platform, targeting various use cases from academics to health-care.
The purpose of this digital twin is to provide an open-source simulator for the Reachy, allowing users to learn how to program a Reachy without having to purchase a physical unit.
The roadmap of this digital twin will focus on the usage of Python and HARFANG® as a fully functional platform for any virtual reality project.

Snooker simulation

A snooker 3D simulation showcasing real-time physics, entirely written in Python 3 using HARFANG® Framework & Bullet Physics engine. The game is rendered with the AAA rendering pipeline with real-time ray-traced reflections without the need of a RTX-capable GPU.
The purpose of this project is both educational and to serve as a platform for reinforcement learning research, thanks for the rich AI ecosystem of the Python language.