Wirewalk Instituteof AI
mob-255 Intermediate Available

ADAS and Automated Driving Safety Assurance: NHTSA, FMVSS and UL 4600

For vehicle-safety compliance, validation and regulatory staff who must file NHTSA reports on time, certify to FMVSS No. 127, defend a UL 4600 safety case and gate ADS releases while AI drafts and a named person decides.

Level
Intermediate
Study time
14 hours
Lessons
18 in 6 modules
Exam
20 questions

About this course

This course follows Norvale Motors, a fictional composite US electric-vehicle maker with a Level 2 hands-free highway feature, and its subsidiary Norvale Autonomy, which runs driverless service in Austin and wants to deploy a vehicle with no steering wheel or pedals. Each module is one problem the team must solve against a clock or a gate: triaging incident signals for Standing General Order 2021-01 reports due five calendar days after notice; finding a defect trend in complaints, warranty claims and field reports after an over-the-air update and briefing the committee that decides whether a Part 573 report is due in five working days; certifying to FMVSS No. 127 before September 1, 2029; preparing a UL 4600 safety case for an independent assessor; gating an ADS release by scenario class and watching safety performance indicators in the field; and building a clause-by-clause Part 555 exemption case modeled on how NHTSA analyzed the Zoox grant, then telling leadership what the 2,500-vehicle limit and NHTSA's review time mean for a promised launch date. Every lesson works a full example on realistic synthetic materials, compares the approaches and tools practitioners actually choose between, shows a realistic failure and its fix, and ends with a faded exercise and a model answer. Every module ends in a Decision Lab, and the course ends with a capstone simulation and a readiness package for the safety board.

The line the course holds is simple. AI extracts, clusters, drafts, proposes and checks. It never decides that a crash is not reportable, never makes or avoids a safety-defect determination, never certifies compliance, never marks a safety-case claim supported and never approves a release or a filing. Each lesson names the person who does. Real-world cases come from the public record: NHTSA's own Federal Register notices, the FMVSS No. 127 text, Part 573 recall reports filed by Waymo and Tesla, NHTSA's SGO audit query to Tesla, the Cruise consent order, Ford's 2024 consent order over recall timeliness as Ford disclosed it, and the Zoox exemption grant. Vendor products (Applied Intuition, Foretellix, MathWorks RoadRunner Scenario, CARLA, dSPACE SIMPHERA, Jama Connect, NVIDIA Halos, Kognic) are described from their own pages as of October 2026 and labeled as vendor claims; Wirewalk does not endorse them. Process standards for the ML component (ISO 26262, ISO 21448, ISO/PAS 8800) are taught in mob-245 and appear here only as evidence sources. Copyrighted standards are described in the course's own words, never reproduced. This course is educational, not legal advice.

What you will be able to do

  • Evaluate incident packets for SGO 2021-01 reportability: quote the source text for each criterion or mark it not stated, reconstruct the timeline from offset EDR, log and video sources to answer the 30-second engagement question, assign the request and due date from first notice, and route every missing fact to a query
  • Judge whether an AI tool is fit for a safety-compliance role from gold-set or agreement results, with an acceptance rule set before the results, false negatives weighted by consequence, the confidence the sample supports, and version pinning and re-test triggers
  • Analyze complaints, warranty claims, field reports and exposure data with AI clustering and de-duplicated, exposure-normalized rates by software version, and write a committee brief that states facts, uncertainty and the open Part 573 determination without recommending an outcome
  • Evaluate an AI-generated FMVSS No. 127 test matrix clause by clause against 49 CFR 571.127, judge results for contact, thin margins and false activation, and decide where simulation can and cannot substitute for track time
  • Evaluate a UL 4600 safety-case fragment by classifying claim-to-evidence links for version and vehicle match, rating AI-proposed defeaters with recorded reasons, and computing ODD coverage against a stated taxonomy with synthetic and real-world evidence counted separately
  • Design a scenario-class release gate and a field safety-performance-indicator monitor that catch a rare, severe regression an improved aggregate hides, and write the go/no-go memo and investigation ticket
  • Evaluate a Part 555 exemption case for an ADS vehicle without manual controls with a clause-by-clause traceability table, rating each equivalence claim against evidence from the deployed configuration and checking the deployment plan and launch date against statutory limits, grant-style conditions and precedent
  • Create a safety readiness package that combines module deliverables, names the decider for each item, states open risks and the honest effect on the business plan, and gives a safety board a recommendation it can act on

Syllabus

6 modules · 18 lessons · about 14 hours

Module 1: The Five-Day Clock: SGO Crash Reporting With AI Extraction

Triage incident signals for Standing General Order reports with evidence-quoted AI extraction and no auto-close, and reconstruct a crash timeline from offset sources.
3 lessons
  1. Who Files What, and When: The US Rulebook Through One Norvale Crash
  2. Triage Without Auto-Close: Extracting SGO Criteria With Evidence Quotes
  3. Was It Engaged? Reconstructing a Crash From EDR, PilotLane Logs and Video
  • Decision Lab · 12 minThe Five-Day ClockDecide whether an auto-closing classifier, a pre-filter or an evidence-quoted tool runs the SGO queue, then handle a motorcycle notice that arrives on a Thursday afternoon.

Module 2: The Drift on Curves: From Field Signals to a Part 573 Decision

Find a defect trend after an over-the-air update in complaints, warranty claims and field reports, measure it by software version, and brief the committee that owns the determination.
3 lessons
  1. Clustering Complaints, Warranty Claims and Field Reports Without Losing the Severe Few
  2. Rates by Software Version: Exposure, Missing Fields and Simpson's Paradox
  3. The Committee Brief, the Five-Working-Day Clock and the Over-the-Air Remedy
  • Decision Lab · 13 minThe Drift on CurvesFind a defect signal in 2,140 field records after an over-the-air update, measure it by software version, brief the committee and meet the five-working-day clock.

Module 3: Certifying to FMVSS No. 127 Before September 1, 2029

Check an AI-built AEB test matrix against the regulation, read results for contact, margins and false activation, and decide how to spend scarce track days.
3 lessons
  1. Checking an AI-Generated AEB Test Matrix Against 49 CFR 571.127
  2. Reading the Runs: Margins, Contact and False Activation
  3. Simulation or Track: Spending Scarce Test Days and Watching the Field
  • Decision Lab · 12 minTrack Days and the 2029 LineDecide what the FMVSS No. 127 program tests, whether to defer track days on a reported lead-time extension, and how to respond to a pedestrian contact and a field signal.

Module 4: A UL 4600 Safety Case for a Vehicle With No Steering Wheel

Prepare a safety case for an independent assessor: version-matched evidence, graded defeaters, honest ODD coverage and data evidence.
3 lessons
  1. Claims, Evidence and Versions: Mapping a GSN Fragment With an LLM
  2. Defeaters and Claim Inflation: Grading an LLM's Safety-Case Suggestions
  3. ODD Coverage and Data as Evidence: Scenario Mining and Label Audits
  • Decision Lab · 13 minBefore the Assessor ArrivesPrepare the K1 safety case for an independent assessor who will sample evidence links himself, while a contradicting test result and a software change arrive.

Module 5: Gating Releases and Watching the Fleet

Gate an ADS release by scenario class, find a safety-indicator shift after a map update, and review remote-assistance events and where the operators sit.
3 lessons
  1. A Release Gate by Scenario Class: When the Aggregate Hides the School Bus
  2. Safety Performance Indicators: Finding a Step Change After a Map Update
  3. Remote Assistance Events: Classification, Hotspots and Where the Operators Sit
  • Decision Lab · 12 minRelease Night for 2.9Decide how NA-ADS 2.9 is gated and shipped when the aggregate improves and one severe class regresses, then answer an indicator alert and a vendor's offshore offer.

Module 6: The Part 555 Application and the Capstone

Build a clause-by-clause exemption case, check the deployment plan against statutory limits and grant conditions, track rulemaking, and assemble the safety readiness package.
3 lessons
  1. Clause by Clause: A Part 555 Traceability Table for the K1
  2. Living Under an Exemption: Operational Authorizations, Limits and the Rulemaking Watch
  3. Capstone: The Norvale Safety Readiness Package
  • Decision Lab · 12 minThe Filing DateDraft and file the K1 Part 555 exemption application against a promised date, when the equivalence argument, a late test finding and the business plan all test the evidence.
  • Capstone simulation · 40 minThe Readiness ReviewOwn the safety calls behind three public promises from October 2026 to June 2027: a reporting tool, a defect, a release, an exemption filing, an assessor visit and a launch date.

Exam and certificate

The exam unlocks once you have completed every lesson and played each Decision Lab to an ending. It draws 20 questions from a larger bank, spread across the modules, with the answer options shuffled for each attempt. You have 40 minutes and need 80% to pass. After a failed attempt you can try again after 24 hours, up to 3 attempts in total. Feedback is given per module, never as the answer key.

Passing earns a certificate of completion and assessment from the Wirewalk Institute of AI with a unique ID that anyone can check at institute.wirewalk.com/verify. It is not an accredited qualification or a vendor certification. How certification works.

The Wirewalk Institute of AI is independent. This course teaches publicly documented products and is not affiliated with or endorsed by Applied Intuition, Foretellix (Foretify), MathWorks RoadRunner Scenario, CARLA (open source), dSPACE SIMPHERA, Jama Connect, NVIDIA Halos, or Kognic. Product and company names are trademarks of their respective owners.

More in AI in Transport, Logistics and Mobility

All 14 courses in this track
mob-001 Free intro Available

AI in Transport, Logistics and Mobility: The One-Hour Briefing

A free one-hour briefing built on ten real transport jobs, not an overview: ranking late loads by predicted ETA, classifying an invoice to HTSUS, coaching from AI dash-cam events, turning aircraft-health alerts into work orders, coding safety reports, acting on machine-vision track findings, watchkeeping with an AI lookout, planning a CII-aware voyage, retesting a perception model and triaging satellite conjunctions. Each lesson shows where AI helps, where the person who signs must decide, and the rule that governs that task.

1 h · 9 lessons Free intro
mob-210 Intermediate Waitlist

Building and Deploying AI in Logistics and Supply Chain

Building and Deploying AI in Logistics and Supply Chain: an intermediate course in the AI in Transport, Logistics and Mobility track. Syllabus in preparation.

h · 0 lessons Join the waitlist
mob-120 Basic Waitlist

AI for Aviation Professionals

AI for Aviation Professionals: a basic course in the AI in Transport, Logistics and Mobility track. Syllabus in preparation.

h · 0 lessons Join the waitlist