OpenApparatus is an open-source organization building tools to procedurally generate navigation environments — floor plans, mazes, and apparatuses — that can be reproduced exactly from a single seed.
The aim is simple: the environment should never be the unstated variable in a methods section.
Spatial environments in behavioral experiments are rarely reusable across studies. Hand-built scenes don't travel between labs; ad-hoc generators rarely publish their seeds. The result is an unspoken gap in otherwise rigorous reporting — the geometry an animal navigated, or a participant walked through, or an agent trained in, is treated as incidental rather than as part of the experimental record.
OpenApparatus treats the environment as a first-class, citable artifact. A small parameter description, plus a seed, fully specifies it; the environment itself is regenerated on demand. Identical inputs always produce an identical environment, on any machine, in any year.
Rodent navigation paradigms · VR wayfinding studies with human subjects · reinforcement-learning agent training · computational-neuroscience simulations · anywhere a spatial environment needs to be specified once and reproduced exactly.
|
The foundation that every other project builds on — and the one you can use directly in your own analysis or simulation pipelines. |
A desktop authoring environment. Design an apparatus visually, preview it live, and export it for use elsewhere — rendering, simulation, or publication figures. |
The Unity integration. Drop it into a Unity project to bring OpenApparatus environments straight into your scenes. |
One specification. Three surfaces. The same environment, every time.
Every repository ships citation metadata, so GitHub renders a Cite this repository button directly on the project page. A formal DOI will accompany the v0.1 release.
Questions, feedback, and use-case stories are welcome in Discussions. Bug reports and feature requests belong on the relevant project repository.