Developers & research
Behaviour you can reproduce.
Robot behaviour is easy to demonstrate once and hard to demonstrate twice. The development environment is built around the second problem.
Environment
What is available to build against.
Petrixia Town simulation
A complete multiplayer world with objects, timed processes and permissions — not a toy grid.
Robot-independent tool interfaces
Write a skill once; run it against a simulated body or a physical one without changing it.
APIs and protocols
A documented command protocol, an actor gateway, and an event stream that records intent against outcome.
Simulated human users
Scripted and interactive participants, so social behaviour can be exercised without recruiting people for every run.
Repeatable scenarios
The world resets before every run, so two runs of the same task are actually comparable.
Physical and virtual testing
The same benchmark suite scores a simulated body and a physical platform.
Evidence
Scored by the world, not by the robot.
A benchmark harness runs whole scenarios end to end and scores them from the world's own records. The world is reset before every run — without that, a run inherits wherever the last one left things, and an object dropped conveniently nearby can make an identical task look eight times cheaper than it is.
| Scenario | Body | Score | Effort | Note |
|---|---|---|---|---|
| salon_visit | jamie | 100.0 | 7 | Multi-step booking flow completed. |
| wash_up | beatriz | 90.4 | 14 | Navigation costs it two blocked steps. |
| examine_newspaper | bruce | 0.0 | 0 | Target never perceived — see limits. |
Honesty constraints
These are what make a score mean anything. The embodiment reaches the world only over public HTTPS, through endpoints limited to what a body could actually perceive. It must never:
- read the world's database;
- read the world's source or map files;
- use the full-world snapshot the browser is given.
Configuration loading refuses a config containing database credentials, or one pointing at localhost — so the honest path is also the easy one.
Known limits, stated plainly
- Navigation only reaches what has been seen. Perception is radius-limited, so a body cannot route to a landmark it has never observed. That is why one scenario scores zero. It is a real robotics problem, correctly surfaced — not a bug papered over.
- No speech channel yet. The agent endpoints carry no chat, so an embodied agent cannot yet hear a person or reply. This is the main blocker for conversational assistance.
- Reasoning is not yet connected. Scripted plans currently stand in for the agent's decisions. Everything below that seam is finished and does not change when the real thing is attached.
Openness
Open where it helps, proprietary where it counts.
Some components are intended to be open — protocols, tool interfaces and the pieces researchers need in order to reproduce a result. Commercially important platform services remain proprietary. The dividing line is drawn so that an external party can verify a claim without being handed the business.