What problem does it solve?
Building a credible physics simulation requires validating it against real-world measurements, but manual comparison is slow and error-prone. This Skill codifies a workflow that anchors Elodin simulations to recorded telemetry and uses elodin monte-carlo campaigns as an automated test harness, so every model change is judged by a 30-run campaign against recorded reality.
Core Features & Use Cases
- Truth Data Vendoring: Import experimental telemetry with provenance, sanity checks, unit-conversion validation, and cross-validation against documented mission events.
- Truth Ghost Replay: Render the recorded vehicle next to the simulated one using a kinematic ghost entity driven by
el.SimulationTick, keeping exports aligned row-for-row.
- Campaign-Based Calibration: Score every run against truth with fit metrics (RMSE, miss distance), narrow
spec.toml parameter ranges around best-fit runs, and iterate until the sim matches reality.
- Use Case: Reconstructing the Apollo lander descent — vendor NASA telemetry, reconstruct the missing horizontal velocity channel from physics, replay the truth ghost in-sim, and run monte-carlo campaigns to calibrate guidance and physics models against the historical record.
Quick Start
Ask the AI to vendor your recorded telemetry as a reference profile and set up an elodin monte-carlo campaign that scores each simulation run against the truth data.
