0.1 How to use this site
How each lesson is structured, and tips for steady progress
Why this site exists
Understanding the safety of automated driving — especially SOTIF (Safety Of The Intended Functionality, ISO 21448) — requires knowledge from several fields: probability and statistics, machine learning, and how AI works. Yet the textbooks for each field are written separately, and it is hard to see why you are learning each piece.
AI Safety Lab connects them through a single question — “How can we tell that an automated vehicle is safe enough?” — and is organized so that beginners can build up their understanding step by step.
Who this site is for
- Students and working professionals interested in the safety of automated driving and AI
- Newcomers to the automotive industry who want to follow discussions about SOTIF and validation
- AI engineers who want to see how their work connects to safety requirements
- Functional safety practitioners who want to add AI and statistical reasoning to their toolkit
Formulas do appear, but each lesson first explains the idea in words and figures, and only then checks it with math. Symbols such as Σ (sigma) and derivatives are reviewed in “0.2 Reading math notation”.
How each lesson is structured
Every lesson follows the same order. Once you are used to it, you can skim for just the parts you need.
| Block | What it contains |
|---|---|
| What you will learn | What you will be able to do after reading (about three points) |
| Build intuition | Get a feel for the idea through everyday examples and concrete numbers |
| Check with math | Turn the intuition into precise language with a minimum of formulas |
| Connection to automated driving | Where this knowledge is used in automated-driving safety |
| Summary | A recap of the key points |
| Quick quiz | Click a question to reveal the answer |
| Further reading | Related lessons, exercise notebooks and references |
The “Connection to automated driving” block appears in every lesson — even the ones on probability and machine learning — so that you never lose sight of what the knowledge is ultimately for.
What the labels mean
- NEW Newly published (shown during the month of publication)
- UPDATE Content revised (shown during the month of the revision; the lesson page shows the date)
- Coming soon Being written (no link yet)
- Japanese only Available in Japanese; English translation in progress
- Advanced A harder lesson. First-time readers can skip it and still move on
- ▶ Video Video lectures (about 5 minutes each, explaining the key points with slides and narration. Links are added as videos are published on YouTube; a dashed mark means coming soon)
Learning environment
All you need is a web browser. For readers who want to try things hands-on, we plan to provide exercise notebooks for Google Colab (a Python environment that runs in the browser). The exercises are optional; the lessons are written so that you can understand them without doing the exercises.
Terms of use
- This site is an educational explainer. Because ISO and other standards are copyrighted, we do not reproduce their text; we only summarize and explain their ideas.
- When applying standards in professional work, always consult the original text and have your work reviewed by internal or external experts.
- The English pages are translated from the Japanese originals. If the two versions differ, the Japanese version takes precedence.
- If you find an error, we would be grateful if you let us know via the contact form.
- When we correct an error, we mark the lesson with “UPDATE” on the home page, show the revision date on the lesson page, and describe the change in the News section.
- Text and figures are licensed under CC BY 4.0. Example credit: “AI Safety Lab (Ken), ‘1.8 Statistics of rare events’, https://www2u.biglobe.ne.jp/~perfect/en/prob/08-rare-events.html (CC BY 4.0)”. If you adapt the material, please say so. Works by others, such as standards and papers, remain subject to their owners’ terms.
About the author
Ken
Automotive Safety / SOTIF / AI Safety
Drawing on professional experience in automotive functional safety, SOTIF and AI safety, I create learning materials for studying the safety of AI and automated driving from the basics.
Please get in touch via the contact form.