On the Future of Flight Software
summary
A clear engineering brief: aviation needs a method to get software failure probability below 10^-9 per flight hour but regulations like DO‑333 give requirements without a practical toolchain. Real-world data points: F‑35 ≈ 8 million lines of code, Boeing Starliner and 737 Max ongoing software fixes, ExoMars Schiaparelli lost to a one‑second IMU glitch that produced a negative altitude. Current verification tools (PVS, nuXmv, SPIN) and SV‑COMP benchmarks (largest C module ≈ 44 KB) do not scale end-to-end, so the talk argues for compositional verification, runtime sanity monitors (e.g., catch negative altitude), and tool integration for certification.
AI-written summary · about
chapters (11)
- 0:00The challenge of flight software
- 5:07Cybersecurity in aerospace
- 10:02Scale and complexity in avionics
- 13:26Gaps in flight regulations
- 17:06Verification tools and SV comp
- 20:35Tool chains and automation
- 23:46Education and maintenance issues
- 26:39The problem of specifications
- 30:10Roadmap for future avionics
- 34:06System integration and security
- 41:46Q&A on future technologies
related talks
15mDiscussion of Aviation Software Oversight Improvement
01h 07mContrasting Embedded Software Development for the Space Shuttle and the Orion MPCV
36mKeynote: The Future of Software 
50mWhen Human Life Depends on Software - Introduction to Safety-Critical Systems
41mSoftware in Space: Lessons Every Developer Can Learn From
42mCode as Risk
58mIs Software Engineering Still an Oxymoron?
44mReflections around SW obsolescence