Analysis & Deep Dives
Data journalism and investigative analysis of autonomous vehicle safety. Every article uses real numbers from the NHTSA incident database.
📊 Our Analysis Methodology
Every analysis article on AutoPilotWatch follows a rigorous, data-first methodology. We start with the raw data — NHTSA SGO incident reports, consumer complaints, recall records, and investigation filings — and apply statistical analysis to identify patterns, trends, and anomalies.
Our analysis principles:
- Data-driven: Every claim is backed by specific numbers from official sources.
- Transparent: We show our work. Data sources and methodology are documented.
- Contextualized: Raw numbers without context are misleading. We always account for fleet size, reporting thresholds, and other confounding factors.
- Independent: We have no financial relationship with any manufacturer. Our analysis is based solely on public data.
For full details on how we process and classify data, see our methodology page. All underlying data is available on our downloads page.
Statistical analysis of NHTSA data, revealing patterns and trends in autonomous vehicle incidents.
Geographic Risk: Your City's AV Safety Score
Not all cities are equal for autonomous vehicle safety. San Francisco has 1,170 incidents while most cities have zero. See the full geographic risk breakdown.
The Cost of AV Crashes: An Insurance Perspective
68 fatalities, 3,096 injuries, 3,051 property-damage incidents — estimated total cost: $195 million. The economic toll of autonomous vehicle crashes.
Where Do AV Crashes Happen? A Roadway Analysis
Streets, highways, intersections, parking lots — we break down 6,215 AV/ADAS crashes by roadway type to find where automated driving fails most often.
Tesla's 3,092-Incident Problem: The Autopilot Crash Record
Tesla accounts for nearly half of all reported AV/ADAS incidents. We break down every number — by model, severity, and trend.
Cruise: From $10 Billion Bet to Total Shutdown
GM's Cruise racked up 155 incidents and then abruptly ceased operations. Here's the full timeline of what went wrong.
California's AV Testing Empire: 2,739 Incidents and Counting
California accounts for 44% of all AV incidents — 1,592 ADS and 1,147 ADAS. San Francisco alone has 1,170 crashes.
AV Crash Trends 2021–2026: The Numbers Are Going Up
From 90 incidents in 2021 to 593 in 2025 — AV crashes are rising every year. Here's the full trend analysis.
Pedestrian Safety and Autonomous Vehicles: The Data We Have
From Uber's fatal Tempe crash to today's 6,215 incidents — what does the data say about pedestrians and AVs?
Night vs Day: When Do Self-Driving Cars Crash Most?
Peak hour: 4 PM with 343 incidents. Quietest: 4 AM. The time-of-day data reveals when autonomous systems are most stressed.
Rain, Snow, and Fog: How Weather Affects AV Safety
3,550 incidents in clear weather vs 432 in rain and just 19 in snow. The counterintuitive truth about weather and autonomous vehicle crashes.
The Speed Factor: Pre-Crash Speed Distribution in AV Incidents
1,756 incidents at 0–25 mph, 613 at 66+ mph. Low-speed crashes dominate but high-speed ones kill. Average: 36 mph.
Beyond Tesla: The 59 Other Companies Reporting AV Crashes
Waymo has 1,729 incidents. GM has 265. Cruise 155. Meet the other companies in NHTSA's autonomous vehicle database.
Speed vs Safety: At What Speed Do AV Crashes Kill?
41% of AV crashes happen under 25 mph, but fatality rates spike above 45 mph. We analyze 6,215 crashes by speed bracket to find the danger zones.
What Crash Reports Actually Say: A Text Analysis
We text-mined 4,218 NHTSA crash narratives to find the most common AV failure patterns. Lane keeping failures lead at 27%, followed by object detection at 23%.
The Acceleration Problem: Are AV Crashes Getting Worse?
593 AV crashes in 2025, up 51% from 2024. But fleet size grew 45%. Is the crash rate actually increasing? We normalize the numbers.
Waymo Expansion 2026: New Cities, 250K Weekly Rides & Safety Milestones
Waymo's 2026 expansion into Miami, Atlanta, Washington DC, and Tokyo. 250K+ weekly paid rides, 170M+ rider-only miles, 92% fewer serious injuries, and the Zeekr RT fleet transition.
Topics We Cover
🔴 Manufacturer Safety
Deep dives into specific manufacturers' safety records, including Tesla, Waymo, Cruise, and emerging AV companies.
📈 Trend Analysis
Year-over-year trends in incidents, fatalities, and injuries. Are autonomous vehicles getting safer over time?
🗺️ Geographic Patterns
Where crashes happen and why. State-level regulation, urban vs. suburban patterns, and climate effects.
⚙️ Technology Comparisons
Camera-only vs. lidar, ADS vs. ADAS, and how different technology approaches affect real-world safety.
📋 Regulatory Analysis
How NHTSA, state regulators, and international bodies are responding to autonomous vehicle safety challenges.
🔍 Incident Investigations
Detailed examinations of specific crashes, including contributing factors, system behavior, and aftermath.
Frequently Asked Questions
What kind of analysis does AutoPilotWatch publish?
We publish 34 articles spanning data deep dives, investigations, comparisons, explainers, and policy analysis. Every article uses real numbers from the NHTSA incident database — we never rely on anecdotal evidence or unverified claims.
How are analysis articles different from the data pages?
Data pages (like Incidents, Manufacturers, States) present raw data in searchable, filterable formats. Analysis articles interpret that data — identifying patterns, testing hypotheses, comparing manufacturers, and drawing evidence-based conclusions that raw tables can't convey.
Can I use AutoPilotWatch analysis in my research or reporting?
Absolutely. All our analysis is based on publicly available NHTSA data and is free to cite. We ask that you credit AutoPilotWatch and link back to the original article. If you're a journalist working on a story, we're happy to provide additional data and context.
How often are new articles published?
We publish new analysis articles as significant data updates, trends, or events warrant. Major NHTSA data releases, recalls, and policy changes often trigger new analysis. Subscribe to our newsletter for notifications.
Explore the Data Behind Our Analysis
Stay Updated on AV Safety
Get notified when we add new data, analysis, and tools. No spam — just safety insights.