Elodin has released an open‑source simulation rig that mirrors the constraints of Anduril’s AI Grand Prix, a $500K autonomous drone race. The stack runs on macOS and Linux, bundles a Rust ECS engine, JAX‑style physics, and a Python bridge to a real‑time Betaflight SITL build. Teams can now write autopilot code against a faithful flight‑controller loop before the official qualifier arrives.

The rig ties three subsystems together with a single post‑step callback: six‑DOF rigid‑body dynamics, multi‑rate IMU/barometer/magnetometer sensors, and a GPU‑rendered 640×360 forward‑facing camera matching the VADR‑TS‑002 spec. An 80‑line UDP bridge synchronises precisely Betaflight’s PID loop at 1 kHz, guaranteeing that any controller tuned here behaves identically on real hardware during flight.

Elodin ships a minimal baseline solver—a simple altitude and position PID—that can clear gates out of the box, and provides scripts for installing dependencies, building Betaflight, and launching the editor. The repository includes an architecture roadmap inviting community PRs, while noting differences such as UDP versus MAVLink and a simplified drag model. Users can start iterating immediately and validate code against the same flight stack used in competition.