Evidence at a glance #
Based on the linked sources. The author's reported result and independent confirmation are separate checks.
- AI involvement
- Reported models: Claude. Tools: Flutter, Rust, MediaCodec, ffmpeg. The writeup below describes the work attributed to AI.
- Reported result
- Partial. This is the directory's source-based outcome classification.
- Independent confirmation
- Not reported.
- Human work and limitations
- The project explicitly avoids geofence defeat and secret key extraction, focusing on structural analysis and interoperability for owned hardware.
Original sources
What was done #
Reverse-engineering and interoperability software enabling direct app-to-bird communication with the DJI Phantom 3 Professional, utilizing a pure-Dart DUML implementation and a Rust control core.
The awto-dji project provides interoperability software to communicate directly with a DJI Phantom 3 Professional. The architecture consists of a Flutter application featuring a pure-Dart DUML codec and a Rust-based direct-control core. Reverse-engineering analysis has focused on structural analysis of the protocol, including telemetry scanning and public-airspace safety gates. The project documents a discrepancy in FPGA command ID numbering, where firmware dispatcher traces do not reconcile cleanly with APK command sets. Recent development includes a live H.264 video decoder, though it remains untested against a real NAL stream from the aircraft. The author reports that video decode issues were traced to a capture-completeness gap rather than a pipeline bug, as the stream lacked SPS/PPS/IDR frames. Additionally, the project investigated DJI Assistant 2 debug mechanisms, finding that debug mode can be accessed via launch arguments, though it is not yet confirmed if this enables P3 support. The project maintains a README and extensive documentation of its reverse-engineering efforts.