San Francisco remains the global epicenter for commercial robotaxi service. If you've searched for San Francisco driverless cars and wondered whether you can actually hail one, how safe they are, and what the experience costs compared to Uber or Lyft, this guide breaks it all down with real numbers and current operational data.
As of early 2026, two primary services operate driverless vehicles on San Francisco streets: Waymo (Alphabet/Google's autonomous driving unit) and Zoox (Amazon's autonomous vehicle subsidiary). Both hold permits from the California Public Utilities Commission (CPUC) and the California Department of Motor Vehicles (DMV) to carry paying passengers without a human behind the wheel. Here's exactly what you need to know.
Is This Truly Self-Driving? SAE Automation Levels Explained
The most common confusion in autonomous vehicles is conflating consumer driver-assist features with actual robotaxi service. They are fundamentally different categories under the SAE International J3016 standard.
SAE Automation Levels — What Matters for Riders
| Level | Classification | Who Drives? | Example |
|---|---|---|---|
| Level 2 | Driver Assistance | Human (always) | Tesla Autopilot/FSD Supervised, GM Super Cruise, Ford BlueCruise |
| Level 3 | Conditional Automation | System (in narrow conditions), human on standby | Mercedes Drive Pilot (limited markets, low speed) |
| Level 4 | High Automation | System (in defined ODD), no human needed | Waymo, Zoox robotaxis in San Francisco |
| Level 5 | Full Automation | System (everywhere, all conditions) | Does not exist commercially |
Key takeaway: San Francisco driverless cars from Waymo and Zoox operate at SAE Level 4 — the vehicle handles all driving tasks within a defined operational design domain (geofenced area, weather conditions, speed range) with no human driver present. This is categorically different from a Tesla on FSD Supervised, where a licensed driver must remain attentive and ready to intervene at all times.
Where Can I Use Robotaxi Services Right Now?
Service areas in San Francisco have expanded considerably since the early pilot days, but they are not unlimited. Both operators use geofencing — virtual boundaries that restrict where their vehicles can pick up, drop off, and navigate.
| Service | San Francisco Coverage | Hours | Vehicle | Other Live Cities (2026) |
|---|---|---|---|---|
| Waymo | Most of SF proper, including downtown, SoMa, Mission, Sunset, Richmond, Marina, and SFO airport routes | 24/7 | Jaguar I-PACE (custom-equipped EV) | Phoenix, Los Angeles, Austin, portions of Bay Area peninsula |
| Zoox | Select zones within SF (SoMa, downtown core, Mission Bay corridor) | Limited hours (typically daytime/evening) | Zoox purpose-built carriage (no steering wheel) | Las Vegas (limited), Foster City |
Important: Waymo's San Francisco service area covers the vast majority of the city as of 2026, including highway segments for airport runs. Zoox operates a more restricted footprint with its distinctive bidirectional carriage vehicle. Both services exclude certain steep hills, construction-heavy corridors, and unmapped private roads.
How to Ride and What It Costs vs. Uber/Lyft
Using a robotaxi in San Francisco is straightforward — download the operator's app, set your pickup and dropoff within the geofenced zone, and a driverless vehicle arrives. There's no human driver to rate or tip.
Pricing Comparison
| Metric | Waymo | UberX (SF) | Lyft Standard (SF) |
|---|---|---|---|
| Base fare | ~$5.00 | ~$3.50 | ~$3.00 |
| Per-mile rate | ~$1.50–$2.00 | ~$1.20–$1.80 | ~$1.10–$1.60 |
| Per-minute rate | ~$0.30–$0.50 | ~$0.25–$0.45 | ~$0.22–$0.40 |
| Typical 5-mile ride | $15–$22 | $13–$20 | $12–$18 |
| Surge pricing | Yes, but generally less volatile | Frequent, can 2–3x | Frequent, can 2–3x |
| Tip expected | No | Yes (15–20%) | Yes (15–20%) |
The bottom line on cost: Waymo pricing is generally comparable to Uber/Lyft before tip, and often slightly cheaper on a total-cost basis once you factor in that no tip is expected. Surge pricing exists but tends to be less extreme because Waymo's fleet capacity is more predictable. Zoox has offered promotional free-ride periods during its expansion phases; standard pricing is expected to be competitive with Waymo once fully commercialized.
How Safe Are San Francisco Driverless Cars? The Real Data
Safety is the most scrutinized aspect of robotaxi operations. California requires operators to report disengagements (moments when the system hands off control or a safety operator intervenes) and collision data to the California DMV and NHTSA.
Waymo Safety Record — Key Figures
- Cumulative fully autonomous miles driven: Over 40 million driverless miles (across all cities) as of late 2025 reporting.
- Collision rate vs. human benchmarks: Waymo's published data (peer-reviewed and submitted to NHTSA) indicates a ~85% reduction in injury-causing crash involvement per million miles compared to human drivers in comparable urban environments.
- Contactless incidents: The majority of reported collisions involve the robotaxi being rear-ended or side-swiped by human-driven vehicles at low speed — consistent with urban stop-and-go traffic patterns.
- SF-specific incidents: San Francisco has seen isolated incidents including a school bus contact (no injuries, bus at fault), a tow-truck encounter, and a few minor fender-benders. No fatalities involving a Waymo vehicle in driverless mode have been reported in SF.
Sources: Waymo safety reports, NHTSA Standing General Order crash data, CPUC incident filings.
Zoox has accumulated fewer public miles given its later commercial launch and more restricted service area. Its purpose-built vehicle includes redundant steering, braking, and sensor systems with no single point of failure. Detailed disengagement and collision data is reported to the DMV but published less frequently than Waymo's.
Context that matters: Urban driving in San Francisco involves dense pedestrian traffic, cyclists, double-parked delivery trucks, and narrow streets. Robotaxis are programmed conservatively — they yield more than human drivers, maintain larger following distances, and will pull over and wait rather than attempt an uncertain maneuver. This conservative behavior occasionally frustrates human drivers behind them, but the data suggests it translates to fewer at-fault collisions.
The Technology Stack: Lidar, Cameras, and Geofencing
Understanding what's on the roof and under the hood of these vehicles helps explain both their capabilities and their limitations.
Waymo (6th-Generation Driver on Jaguar I-PACE)
- Lidar: Multiple roof-mounted and body-integrated lidar units providing 360° 3D point-cloud mapping out to 500+ meters. Lidar works independently of lighting conditions.
- Cameras: High-resolution camera array (8+ cameras) for object classification, traffic-light state detection, and sign reading.
- Radar: Imaging radar for velocity detection and performance in rain/fog where cameras degrade.
- HD Maps: Pre-built centimeter-accuracy maps of every street in the service area. The vehicle localizes within this map in real time.
- Compute: Custom AI inference hardware processing sensor fusion at ~10 Hz+ decision cycles.
Zoox (Purpose-Built Carriage)
- Sensor architecture: Similar lidar-camera-radar triad, but packaged into a symmetric vehicle with no designated "front" — the carriage can drive equally in either direction.
- Redundancy: Dual-redundant compute, steering, and braking. If one system fails, the backup takes over without passenger intervention.
- No steering wheel: The Zoox carriage has no manual driving controls — a design choice that requires Level 4 reliability from the start.
Lidar vs. camera-only debate: Tesla's consumer FSD system uses a camera-only ("Tesla Vision") approach, arguing that neural networks can extract sufficient 3D understanding from 2D imagery. Waymo and Zoox both rely on lidar as a primary sensor, arguing that direct depth measurement provides a safety-critical redundancy that cameras alone cannot guarantee — especially in low light, glare, or occlusion scenarios. In commercial robotaxi operations, the lidar-inclusive approach currently dominates.
Geofencing: Both services operate within a pre-mapped Operational Design Domain (ODD). The vehicle knows every lane, crosswalk, speed limit, and signalized intersection within its territory. If you request a ride outside the geofence, the app simply won't allow it. This is a deliberate safety boundary, not a software limitation that can be patched over-the-air overnight.
The Regulatory Landscape: Who Approves Robotaxi Operations?
Operating a commercial driverless service in California requires approval from multiple agencies, and the regulatory environment has tightened since the high-profile Cruise incidents in late 2023 (which led to Cruise suspending all operations and eventually being shuttered by GM in 2024).
- California DMV: Issues the Autonomous Vehicle Testing and Deployment permits. Operators must submit disengagement reports, collision reports, and proof of insurance.
- California Public Utilities Commission (CPUC): Regulates the commercial passenger-service aspect — essentially the "taxi license" for robotaxis. The CPUC reviews safety data, accessibility plans, and equity commitments before granting fare-charging authority.
- NHTSA (federal): Oversees vehicle safety standards. Current FMVSS (Federal Motor Vehicle Safety Standards) were written for human-driven cars, and Zoox received a specific exemption to operate its steering-wheel-free carriage. NHTSA also collects crash data through its Standing General Order for automated driving systems.
- San Francisco Municipal Transportation Agency (SFMTA): Manages street-level operations, curb access, and coordinates with operators on construction zones and event closures.
The post-Cruise regulatory environment means both Waymo and Zoox face more scrutiny on incident reporting, emergency-response interaction, and public transparency than they did in 2022. This is net-positive for rider and pedestrian safety, though it slows expansion timelines.
Current Limitations: What Robotaxis Cannot Do
Honest Limitations of SF Driverless Cars in 2026
- Heavy rain and fog: While lidar performs better than cameras in low visibility, extreme weather still degrades sensor performance. Waymo may reduce speed, increase following distance, or temporarily restrict service during heavy atmospheric rivers.
- Construction zones: Unexpected lane closures, temporary barriers, and flaggers directing traffic remain challenging edge cases. Vehicles may reroute around construction rather than navigate through it.
- Unmapped areas: If a destination falls outside the geofence — including certain parking garages, private driveways, or newly built developments not yet in the HD map — the vehicle cannot complete the trip.
- Parking and pickup precision: Robotaxis pull to the curb at designated or safe stopping points. They cannot enter parking structures, valet themselves, or navigate gated communities.
- Emergency vehicles: While systems detect sirens and pull over, complex emergency scenes (fire trucks blocking multiple lanes, police directing traffic) can cause the vehicle to stop and wait until the path is clear.
- Child seats and accessibility: Waymo's I-PACE fleet does not offer child car seats. Wheelchair-accessible vehicles remain limited, though both operators have committed to ADA-compliant options as fleets expand.
- No intercity trips: You cannot take a Waymo from San Francisco to Lake Tahoe or Sacramento. Highway operation exists for specific routes (e.g., SFO airport runs), but long-distance highway driving is not part of the current ODD.
What the Experience Is Actually Like
For first-time riders, the most striking aspect is the absence of a driver. You unlock the vehicle through the app (typically by tapping your phone on a reader at the door handle), buckle up, and confirm your ride on a rear-seat screen. The vehicle then executes the route autonomously — stopping at signals, yielding to pedestrians, changing lanes, and navigating around double-parked cars.
The driving style is deliberately cautious. Expect slightly longer trip times than an aggressive human Uber driver would deliver — Waymo vehicles maintain generous following distances, come to full stops at stop signs (never a "California roll"), and will not attempt yellow-light dashes. For many riders, this conservative approach is reassuring; for those in a rush, it can feel overly deliberate.
The in-cabin experience includes climate control, a screen showing the vehicle's perception of surrounding objects (cars, cyclists, pedestrians rendered in real time), music controls, and a "pull over" button if you need to end the ride early. Support is available via in-app chat or a call button that connects to a remote assistance team — these operators do not drive the vehicle remotely but can advise the system on how to proceed in ambiguous situations.
Frequently Asked Questions
Do I need a special account or waitlist to ride a driverless car in San Francisco?
Waymo's service is open to the general public through the Waymo One app — no waitlist as of 2026. Simply download the app, create an account, and request a ride within the service area. Zoox availability may still require an invite or be limited to certain zones and hours during its expansion phase.
Can a robotaxi take me to SFO or Oakland Airport?
Waymo offers service to San Francisco International Airport (SFO) from within its SF geofence, including highway segments. Service to Oakland Airport (OAK) depends on whether your origin falls within the current East Bay/Peninsula service boundary — check the app for real-time availability.
What happens if the vehicle encounters a situation it can't handle?
The vehicle will come to a safe stop (typically pulling to the curb) and contact a remote assistance team. The remote operator does not take direct control of steering or braking but can advise the system — for example, confirming that a construction barrier is temporary and the vehicle should proceed around it. If the situation cannot be resolved, the ride may be canceled and you'll be refunded.
Are robotaxis insured?
Yes. Both Waymo and Zoox carry commercial insurance policies that cover passengers, pedestrians, and property damage during autonomous operation. California requires minimum coverage levels as a condition of the CPUC permit. As a rider, you do not need separate insurance to use the service.
How do San Francisco driverless cars compare to Tesla's FSD?
They are fundamentally different products. Waymo and Zoox are SAE Level 4 services — no human driver is present or needed within their operating domain. Tesla's Full Self-Driving (Supervised) is SAE Level 2 driver-assist — the human in the driver's seat is legally responsible for the vehicle at all times and must remain attentive. Tesla FSD is a consumer feature you use in your own car; Waymo and Zoox are on-demand ride services. Comparing them directly is like comparing a personal gym membership to a personal trainer — related domain, entirely different delivery model and responsibility structure.
Will robotaxi service expand beyond San Francisco soon?
Waymo already operates commercially in Phoenix, Los Angeles, and Austin, with continued expansion planned for additional metro areas. Zoox is growing its footprint in Las Vegas and the Bay Area. However, each new city requires extensive mapping, regulatory approval, and local fleet deployment — expansion is measured in years, not months. San Francisco remains the densest, most mature robotaxi market in the world.
