Metriplane

Metriplane

Metriplane is an open-source Python system for observing bounded robotic and industrial workcells and turning observations into reproducible technical evidence.

It converts camera observations into metric planar object state and provides tooling for recording, replaying, evaluating, and verifying that state.

Core capabilities

• Camera-based planar state observation
• Metric mapping and calibration
• Multi-camera fusion and object tracking
• Deterministic record/replay
• Zone dwell and transition analysis
• Incident detection and evidence bundles
• Provenance and configuration tracking
• Regression testing and reproducibility checks
• Bounded-queue and overload/backpressure evaluation
• CPU/GPU equivalence testing
• ROS 2 bridge for publishing Metriplane state, alerts, and incidents

Metriplane is designed for reproducible evaluation of physical workcells rather than robot control. It is observe-only and can be used to preserve evidence from experiments, reproduce recorded runs, and detect behavioral regressions.

Getting started

Documentation, examples, and installation instructions are available on the Metriplane website and GitHub repository.

• Website: https://www.metriplane.com/
• Source code: https://github.com/Miko997/metriplane
• PyPI: https://pypi.org/project/metriplane/
• Software archive: https://archive.softwareheritage.org/