If you search for self-driving cars statistics, you will find two parallel universes. One is full of marketing superlatives and vague promises about the future of mobility. The other is buried in regulatory filings, disengagement reports, and incident databases maintained by agencies like the National Highway Traffic Safety Administration and state regulators. This guide lives in the second universe.
As of early 2026, true driverless robotaxi services — operating without a human behind the wheel — are a reality, but only in a handful of U.S. cities. Meanwhile, consumer vehicles with advanced driver-assistance systems remain firmly at SAE Level 2, meaning the driver must stay engaged at all times. Conflating these two categories is the single most common error in autonomous vehicle coverage. Below, we separate them with hard numbers.
Understanding Autonomy Levels: Level 2 vs. Level 4
Before parsing any statistic, you need to understand what level of automation you are looking at. The Society of Automotive Engineers (SAE) defines six levels of driving automation, and the gap between Level 2 and Level 4 is not incremental — it is categorical.
| Level | Name | Who Drives? | Examples | Liability |
|---|---|---|---|---|
| Level 0 | No Automation | Human | Basic cruise control | Driver |
| Level 1 | Driver Assistance | Human (system assists steering OR speed) | Adaptive cruise, lane-keep | Driver |
| Level 2 | Partial Automation | Human (system assists steering AND speed simultaneously) | Tesla Autopilot, GM Super Cruise, Ford BlueCruise, Tesla FSD (Supervised) | Driver |
| Level 3 | Conditional Automation | System (in specific conditions; human must take over on request) | Mercedes DRIVE PILOT (limited markets, low-speed highway) | Shared/conditional |
| Level 4 | High Automation | System (no human needed in defined operational domain) | Waymo One, Zoox | Operator/company |
| Level 5 | Full Automation | System (everywhere, all conditions) | Does not exist commercially | N/A |
Where Can You Actually Use a Robotaxi Right Now?
Level 4 robotaxi availability in the U.S. remains geographically constrained. Here is where commercial, driverless service operates as of early 2026:
| Operator | City | Service Area | Hours | Vehicle |
|---|---|---|---|---|
| Waymo One | Phoenix, AZ | ~225 sq mi (metro Phoenix including Chandler, Tempe, Mesa, parts of Scottsdale) | 24/7 | Jaguar I-PACE (Gen 5 sensor suite) |
| Waymo One | San Francisco, CA | ~50 sq mi (most of SF proper, expanded to include SFO airport trips) | 24/7 | Jaguar I-PACE (Gen 5) |
| Waymo One | Los Angeles, CA | ~80 sq mi (Santa Monica, Hollywood, downtown LA, expanding) | Most hours | Jaguar I-PACE (Gen 5) |
| Waymo One | Austin, TX | Limited initial deployment (downtown and south Austin) | Limited hours | Jaguar I-PACE (Gen 5) |
| Zoox | Las Vegas, NV | Limited corridor (Las Vegas Strip area) | Limited hours | Zoox purpose-built vehicle (no steering wheel) |
| Zoox | Foster City, CA | Testing/limited service | Limited | Zoox purpose-built vehicle |
If you live outside these metro areas, you cannot currently hail a driverless robotaxi. Announced expansion plans — Waymo has discussed additional cities — should not be confused with live service. Check the operator's app for your specific address before assuming coverage.
Self-Driving Cars Statistics: Safety Data in Context
Safety is the metric that matters most, and it is the hardest to evaluate honestly. Here is what the data actually shows.
• Over 50 million fully autonomous (driverless) miles completed across Phoenix, San Francisco, and Los Angeles.
• Waymo's published analysis (peer-reviewed methodology) indicates a ~85% reduction in injury-causing crash rates compared to human-driven benchmarks in the same operating areas.
• Contact-rate incidents (any physical contact, including minor ones at low speed) reported at roughly ~5.3 per million miles for driverless operations, vs. a human benchmark of ~20+ per million miles in comparable urban driving.
• The vast majority of robotaxi contact events occur at very low speeds (under 10 mph) and involve other road users (pedestrians, cyclists, human drivers) making unpredictable moves.
What these numbers mean — and what they do not:
- Geofencing matters enormously. Waymo operates in mapped, well-understood urban environments with favorable weather. These statistics do not transfer to unmapped rural roads, heavy snow, or construction zones with altered lane markings.
- The comparison baseline is critical. Urban driving with frequent stops, intersections, and pedestrians is inherently higher-risk than highway cruising. Comparing robotaxi urban crash rates to the national average (which includes vast amounts of low-risk rural highway driving) would be misleading.
- NHTSA Standing General Order data requires all Level 2+ and Level 4 operators to report crashes. This database, accessible via nhtsa.gov, provides raw incident counts but requires careful normalization by miles driven to be meaningful.
- Cruise's 2023 suspension in California remains an instructive case. After a serious incident in San Francisco where a pedestrian was dragged, California DMV suspended Cruise's permit. Cruise (GM's autonomous unit) subsequently restructured operations with significantly reduced scope. This demonstrates that safety incidents can halt operations entirely — the technology is not infallible.
For the most current crash reporting data from all automated driving systems, refer to the NHTSA Automated Vehicles Safety page.
Robotaxi Pricing vs. Uber and Lyft: The Real Cost
One of the most practical self-driving cars statistics is what a ride actually costs. Robotaxi pricing has been competitive with rideshare, often subsidized during expansion phases.
| Metric | Waymo One | UberX (Comparable) | Lyft (Comparable) |
|---|---|---|---|
| Base fare (typical) | $3.50–$5.00 | $2.50–$4.00 | $2.50–$4.00 |
| Per-mile rate | $1.25–$1.75 | $1.00–$1.50 | $1.00–$1.50 |
| Per-minute rate | $0.25–$0.40 | $0.20–$0.35 | $0.20–$0.35 |
| 5-mile, 15-min trip estimate | $13–$20 | $11–$17 | $11–$17 |
| Surge pricing | Less volatile (fixed fleet) | Highly variable | Highly variable |
| Tipping | Not expected (no driver) | Expected ($2–$5+) | Expected ($2–$5+) |
| Promotional discounts | Frequent for new users | Frequent | Frequent |
Key takeaways on cost:
- On a per-ride basis, Waymo is typically within 10–20% of Uber/Lyft pricing for comparable trips — sometimes cheaper, sometimes slightly more expensive.
- When you factor in that no tip is expected on a robotaxi ride, the effective cost is often lower than rideshare for the same distance.
- Robotaxi pricing is less susceptible to surge multipliers because the fleet size is fixed and managed by the operator, though wait times may increase during peak demand.
- Promotional pricing for new riders can make initial rides significantly cheaper — Waymo has offered first-ride credits in new markets.
How to Use a Robotaxi Service
Using Waymo One is nearly identical to using Uber or Lyft: download the app, enter your destination, confirm pickup, and a vehicle arrives. The difference is that no one is in the driver's seat. You unlock the car via the app (or by tapping your phone on the vehicle), select your music and climate preferences on the rear screen, and the vehicle navigates to your destination. A support line is accessible via in-car button at all times.
Technology Stack: How Robotaxis Actually See the Road
Understanding the self-driving cars statistics requires understanding the hardware and software generating them. Level 4 robotaxis use a fundamentally different sensor approach than consumer Level 2 vehicles.
Waymo's Gen 5 Sensor Suite
- LiDAR: Multiple custom-built lidar units providing 360-degree, high-resolution 3D point clouds. LiDAR works independently of ambient light and provides precise distance measurement to objects within ~500 meters.
- Cameras: High-resolution camera array covering 360 degrees for object classification (distinguishing a pedestrian from a cyclist from a trash can), traffic light state detection, and sign reading.
- Radar: Multiple radar units for velocity detection of moving objects and performance in adverse conditions (rain, fog) where cameras and lidar degrade.
- HD Maps: Centimeter-precision pre-mapped routes. The vehicle operates only within its mapped geofence and uses the map as a reference layer for localization and route planning.
- Compute: Custom onboard computing platform processing sensor fusion, perception, prediction, and planning in real time.
Camera-Only vs. Multi-Sensor: The Debate
Tesla's consumer FSD (Supervised) system relies exclusively on cameras (no lidar, no radar in current hardware). This is a viable approach for Level 2 driver-assist where a human remains responsible. However, no Level 4 commercial robotaxi operator uses cameras alone. Every current Level 4 service uses lidar as a core sensor. The redundancy of multiple sensor modalities (lidar + camera + radar) is considered essential for operations where no human fallback exists.
Zoox's Purpose-Built Approach
Zoox (owned by Amazon) takes a different path: a custom-designed vehicle with no steering wheel or pedals, built from the ground up for autonomous operation. The sensor suite is integrated into the vehicle body rather than retrofitted. Zoox vehicles are bidirectional (can drive equally well in either direction) and feature four-wheel steering. This design eliminates the need to retrofit a consumer vehicle but limits production scale.
Regulatory Landscape: Who Approves Robotaxi Operations?
Robotaxi deployment in the United States is governed by a patchwork of state and federal regulations — there is no single national permit.
Key Regulatory Bodies
- State DMVs: Most states require a permit or deployment license for driverless testing and commercial operations. California's DMV, for example, issues testing permits and deployment permits with specific conditions. Arizona has taken a more permissive approach, which is why Phoenix was Waymo's first commercial market.
- State Public Utility Commissions (PUCs): In California, the CPUC (California Public Utilities Commission) must authorize commercial passenger service — this is a separate approval from the DMV permit.
- NHTSA (Federal): Sets vehicle safety standards (FMVSS) and requires crash reporting via Standing General Order for all ADS-equipped vehicles. NHTSA does not currently "approve" robotaxi services but investigates incidents and can issue recalls or enforcement actions.
- Local governments: Cities can impose operational restrictions, including geofencing requirements, hours-of-operation limits, and fleet-size caps. San Francisco's Board of Supervisors has actively debated robotaxi expansion conditions.
The regulatory landscape is evolving. For current state-level requirements, the National Conference of State Legislatures (NCSL) autonomous vehicles tracker maintains a comprehensive database of state legislation.
Current Limitations: Where Robotaxis Cannot Operate
The honest self-driving cars statistics include the scenarios where these systems simply do not work. Level 4 robotaxis are impressive within their operational design domain (ODD), but that domain has hard boundaries.
- Weather: Heavy rain, snow, ice, and dense fog degrade sensor performance (especially lidar and cameras). Robotaxi services reduce operations or suspend them entirely in severe weather. Phoenix's dry climate is not coincidental.
- Geofencing: Vehicles will not operate outside their pre-mapped service area. If your trip crosses the geofence boundary, the service will not accept the ride.
- Construction zones: Major road construction with altered lane markings, temporary barriers, or changed traffic patterns can cause the vehicle to stop and request remote assistance or reroute.
- Complex unpredictable scenarios: Dense crowds, unusual road configurations, emergency situations with first responders directing traffic outside normal patterns — these edge cases can cause the vehicle to pull over safely and wait.
- Specific pickup/dropoff: The vehicle may not be able to pull into certain driveways, parking garages, or private property. Dropoffs occur at mapped, accessible curb locations.
- No highway operations (most services): Current robotaxi services primarily operate on surface streets. High-speed highway driving adds complexity that most operators have not yet commercially deployed.
- Passenger limitations: Car seats for small children, wheelchair accessibility, and large cargo are generally not supported by current robotaxi fleets.
Level 2 Driver-Assist: A Different Category Entirely
Because consumer interest in self-driving cars statistics often stems from considering a vehicle purchase, it is essential to address what you can actually buy today.
Tesla FSD (Supervised), GM Super Cruise, Ford BlueCruise, Mercedes DRIVE PILOT, and similar systems are not robotaxis and are not self-driving in the Level 4 sense.
- Tesla FSD (Supervised): Level 2. Requires constant driver attention. Works on a wide range of roads but the driver is legally liable for all driving decisions. Camera-only. ~$8,000 add-on or subscription.
- GM Super Cruise: Level 2. Hands-free on pre-mapped divided highways (~750,000+ miles of compatible roads in North America). Driver attention monitoring via camera. Included on select trims or available as a package.
- Ford BlueCruise: Level 2. Hands-free on pre-mapped divided highways (~130,000+ miles). Driver monitoring required. Available on select Ford and Lincoln models.
- Mercedes DRIVE PILOT: The closest thing to Level 3 in the U.S. — approved for conditional hands-off driving at speeds up to 40 mph on specific highway sections (Nevada, limited California deployment). Mercedes accepts liability during DRIVE PILOT engagement under specific conditions. Extremely limited availability and high cost.
The safety statistics for these Level 2 systems are tracked separately from Level 4 robotaxi data. NHTSA's Standing General Order crash reporting includes both categories, but the operational contexts are so different that combining them produces misleading conclusions.
Frequently Asked Questions
Where can I use a self-driving robotaxi right now?
Commercial driverless robotaxi service is available in Phoenix, San Francisco, Los Angeles (Waymo One), and limited areas of Austin, TX (Waymo One), plus Las Vegas (Zoox, limited). If you are outside these metros, the service is not available to you regardless of announcements about future expansion.
How much does a robotaxi cost compared to Uber or Lyft?
Expect to pay within 10–20% of standard Uber/Lyft pricing for a comparable trip. A typical 5-mile urban ride costs $13–$20 on Waymo One. The effective cost is often lower than rideshare because no tip is expected and surge pricing is less volatile.
Are self-driving cars safer than human drivers?
Within their operational domain (mapped urban areas, fair weather, moderate speeds), Waymo's published data shows approximately an 85% reduction in injury-causing crash rates compared to human benchmarks in similar driving environments. However, this comparison is specific to the geofenced urban contexts where robotaxis operate. These statistics do not apply to highway driving, rural roads, or adverse weather — scenarios where robotaxis do not currently operate commercially.
What are the main limitations of current robotaxi technology?
Heavy rain, snow, and fog. Strict geofencing to pre-mapped areas. Difficulty with active construction zones and unusual traffic scenarios. No support for car seats, wheelchairs, or large cargo. Most services do not operate on highways. If your trip falls outside these boundaries, the service will not accept the ride.
Is Tesla FSD self-driving?
No. Tesla Full Self-Driving (Supervised) is a Level 2 driver-assistance system. The driver must remain attentive and is legally responsible for all vehicle operation at all times. The "Full Self-Driving" name is a product label, not a description of SAE Level 4 or 5 capability. Tesla's own documentation states the driver must supervise the system and keep hands on or near the wheel.
When will self-driving cars be available everywhere?
There is no credible timeline for nationwide Level 4 or Level 5 deployment. Current operators are expanding city-by-city, and each new market requires extensive mapping, regulatory approval, and operational validation. Industry experts generally expect gradual metro-by-metro expansion over the next decade, with rural and adverse-weather operations significantly further out. Be skeptical of any source promising full autonomy by a specific year.
How many autonomous miles have been driven commercially?
Waymo has completed over 50 million fully driverless (no safety driver) miles across its commercial markets as of late 2025 reporting. This is the largest verified dataset of Level 4 commercial operations in the U.S. Other operators (Zoox, Cruise restructured operations) have significantly fewer commercial driverless miles. For context, the U.S. driving public covers roughly 3.2 trillion miles per year — robotaxi operations represent a tiny fraction of total vehicle miles traveled.
