Reproduce “Mission (Im)Possible: Towards Autonomous Discovery of Autopilot Logic Flaws in Open-Source Drone Frameworks” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Ziyi Liang, Zichuan Li, Bin Yuan, Deqing Zou, Luyi Xing
The authors introduce IMF to find unsafe logic flaws in autopilots by tracing MAVLink commands through code, testing auditor-proposed hypotheses in software-in-the-loop simulation, and checking logs for predicted behavior. On PX4 and ArduPilot, experts verified 105 unique flaws. On a 50-log test, log-analysis F1 was 0.81 with hypotheses versus 0.67 without; results are bounded to these systems and tested scenarios.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.