If you have shopped for a Tesla in the last few years, you have likely encountered the $8,000 option called Full Self-Driving Capability (or its current subscription equivalent). The name implies a vehicle that drives itself. The reality, as of 2026, is more nuanced: Tesla's FSD Supervised remains a Level 2 driver-assistance system that requires your constant attention, while true driverless robotaxis from Waymo and others operate at Level 4 — but only in specific geofenced cities. Understanding the gap between marketing language and engineering reality is essential before you spend money on either a software package or a robotaxi ride.
The SAE Autonomy Scale: Where Full Self Driving Capability Actually Sits
The Society of Automotive Engineers defines six levels of driving automation, from Level 0 (no assistance) to Level 5 (full automation everywhere, no steering wheel needed). This framework is the only reliable way to cut through marketing claims and understand what a system can and cannot do.
| Level | Name | Who Drives? | Who Monitors? | Examples |
|---|---|---|---|---|
| 0 | No Automation | Human | Human | Legacy vehicles |
| 1 | Driver Assistance | Human + system (one task) | Human | Adaptive cruise, lane-keep |
| 2 | Partial Automation | System (steering + speed) | Human (must supervise) | Tesla FSD Supervised, GM Super Cruise, Ford BlueCruise |
| 3 | Conditional Automation | System (in limited conditions) | System (human on standby) | Mercedes Drive Pilot (limited markets) |
| 4 | High Automation | System (geofenced areas) | System (no human needed) | Waymo One, Zoox (pilot programs) |
| 5 | Full Automation | System (everywhere) | System | Does not exist commercially |
Where Robotaxi Services Actually Operate Right Now
Unlike consumer-purchased driver-assist systems that work on most roads, Level 4 robotaxis are restricted to mapped, approved geofenced zones. Here is where you can hail a driverless ride in 2026:
| Service | Operator | Cities | Hours | Vehicle |
|---|---|---|---|---|
| Waymo One | Waymo (Alphabet) | Phoenix metro, San Francisco, Los Angeles, Austin (limited) | 24/7 in most zones | Jaguar I-PACE (custom sensor suite) |
| Zoox | Zoox (Amazon) | Las Vegas (Strip corridor), Foster City, CA (pilot) | Limited hours | Zoox Gen 3 (purpose-built, no steering wheel) |
| Cruise | Cruise (GM) | Phoenix (limited relaunch), select test markets | Daytime only, phased | Chevrolet Bolt EV / Cruise Origin (announced) |
| Motional | Motional (Hyundai/Aptiv) | Las Vegas (Uber/Lyft integration, limited) | Select hours | Hyundai Ioniq 5 (sensor-equipped) |
Expansion announcements are frequent, but only the cities listed above have live, paying-passenger driverless service as of early 2026. If a service says "coming soon" to your city, it is not yet available — regulatory approval, mapping, and safety validation take months per jurisdiction.
Technology Stack: Cameras vs. Lidar and Why It Matters
The sensor architecture behind autonomous driving is a major differentiator between consumer driver-assist and commercial robotaxi platforms.
Camera-Only Approach (Tesla FSD Supervised)
Tesla removed radar and ultrasonic sensors from its vehicles, relying entirely on a suite of 8 cameras and neural-network processing. The philosophy is that cameras approximate human vision, and AI can interpret scenes as a human would. Limitation: Camera-only systems can struggle with depth perception in low-light, heavy rain, fog, and direct sun glare — conditions where ranging sensors excel.
Multi-Sensor Fusion (Waymo, Zoox, Motional)
Level 4 robotaxis combine lidar (light detection and ranging — laser-based 3D mapping), radar (speed/distance in poor visibility), and cameras (object classification, signage, lane markings). Waymo's 5th-generation hardware includes roof-mounted lidar with 360° coverage, corner radars, and 29 cameras. Advantage: Redundant sensing provides reliable perception in rain, fog, and darkness — critical for driverless operation where no human can intervene.
Geofencing
Robotaxis operate within HD-mapped zones (centimeter-precision maps updated regularly). The system knows every lane, curb, signal, and crosswalk in advance. If the vehicle exits the geofence or encounters unmapped construction, it will pull over safely rather than proceed uncertainly. Consumer Level 2 systems have no geofence — they work (with varying competence) on virtually any road, which is both their strength and their risk.
Safety Record: What the Data Actually Shows
Safety is the most contested question in autonomous driving. Here is what verified data tells us as of 2026:
Tesla FSD Supervised data is reported differently. Tesla's quarterly Vehicle Safety Reports cite crash rates per million miles driven with FSD engaged, comparing them to the U.S. average. Critics note that Tesla's data does not isolate FSD from Autosteer, does not control for road type or driver demographics, and relies on self-reporting. The National Highway Traffic Safety Administration (NHTSA) has investigated multiple incidents involving Tesla driver-assist systems, resulting in recalls for software updates related to intersection behavior and driver-monitoring enforcement. As of 2026, FSD Supervised requires camera-based driver monitoring — if you look away from the road, the system escalates warnings and eventually disengages.
Cruise's 2023 incident in San Francisco — where a pedestrian was struck by a human-driven vehicle and then dragged by a Cruise robotaxi that failed to properly detect the person in the roadway — led to the company's California permit suspension and a nationwide pause. Cruise has since relaunched with phased, supervised operations in limited markets. This incident underscores that Level 4 systems, while statistically safer in aggregate, can fail in edge cases that human drivers might handle more gracefully.
Pricing: Robotaxi vs. Rideshare vs. Owning
| Option | Typical Cost (5-mile urban trip) | Per-Mile Estimate | Notes |
|---|---|---|---|
| Waymo One | $12–$18 | $2.40–$3.60/mi | Comparable to UberX; surge pricing during peak; no tip expected |
| UberX / Lyft | $11–$20 | $2.20–$4.00/mi | Varies by surge, driver availability, tip (15-20% customary) |
| Zoox (Las Vegas) | Free–$15 (pilot pricing) | Varies | Limited commercial availability; promotional rates |
| Personal EV (e.g., Tesla Model 3) | $0.60–$1.20 (electricity) | $0.12–$0.24/mi | Excludes purchase, insurance, maintenance, parking; electricity only |
Robotaxi pricing is currently comparable to rideshare — sometimes slightly cheaper because there is no driver to pay, sometimes more expensive due to operational costs of the sensor suite and remote-assistance infrastructure. Waymo has introduced loyalty pricing and subscription plans in some markets. The long-term economic thesis is that removing the driver (who accounts for roughly 60-70% of rideshare costs) will eventually make robotaxis cheaper than human-driven rideshare, but that inflection point has not yet been reached at scale.
Regulatory Landscape: Who Approves Robotaxi Operations?
Autonomous vehicle regulation in the U.S. is a patchwork of state and federal authority:
- Federal (NHTSA): Sets vehicle safety standards (FMVSS). NHTSA can issue exemptions for vehicles without steering wheels or pedals (relevant to Zoox's purpose-built vehicle). NHTSA also investigates crashes and issues recalls.
- State DMVs and Public Utility Commissions: California's DMV issues testing and deployment permits. The California Public Utilities Commission (CPUC) regulates commercial passenger service. Arizona takes a permissive approach — Waymo launched commercially in Phoenix with relatively light state oversight. Texas, Nevada, and other states have their own frameworks.
- Local municipalities: Cities can impose operational restrictions (hours, zones, speed limits) even within state-approved frameworks. San Francisco's Board of Supervisors has pushed for tighter restrictions on robotaxi operations following incidents.
For consumers, this means robotaxi availability depends entirely on your state and city government's willingness to approve commercial operations. There is no federal "robotaxi license" that allows nationwide deployment.
Current Limitations: What Robotaxis and FSD Cannot Do
Hard Limitations — Do Not Expect These Systems to Handle:
- Severe weather: Heavy rain, snow, fog, and ice degrade sensor performance. Waymo reduces speed or suspends service in heavy rain. Tesla FSD may disengage or make errors in low-visibility conditions.
- Unmapped or rapidly changing construction zones: Robotaxis will pull over and request remote assistance or end the ride. FSD may attempt to navigate but can make unsafe lane-choice decisions.
- Complex unprotected left turns in dense traffic: A known challenge for all systems. Robotaxis may wait excessively long or, in rare cases, misjudge gaps.
- Off-road, parking garages, or private property: Geofenced services do not operate outside their zone. FSD can navigate some parking lots but is unreliable in tight or complex structures.
- Emergency vehicle interaction: While systems are programmed to yield, real-world incidents have shown delayed or incorrect responses to approaching emergency vehicles.
- Unusual road users: Horse-drawn carriages, oversized loads, people in costumes, or animals on the road can confuse object-classification models.
The fundamental limitation is that Level 2 systems (including Tesla FSD) require a human who can intervene within seconds, while Level 4 systems handle all situations within their domain but cannot operate outside it. Neither achieves the flexibility of a competent human driver across all conditions.
Decision Framework: Should You Pay for FSD or Use Robotaxis?
Choose Tesla FSD Supervised (or equivalent Level 2) if:
- You drive frequently on highways and want reduced fatigue on long commutes
- You understand you are the driver and must remain attentive
- You live outside robotaxi service areas and need personal vehicle autonomy
- You accept the $8,000 upfront cost or ~$99/month subscription as a convenience feature, not a self-driving solution
Choose robotaxi services (Waymo, etc.) if:
- You live in or visit Phoenix, San Francisco, Los Angeles, or Austin
- You want a genuinely driverless experience with no attention requirement
- You primarily take short-to-medium urban trips where robotaxi pricing competes with rideshare
- You value the safety record of multi-sensor Level 4 systems over the flexibility of a personal vehicle
Do not purchase FSD expecting: Level 4 or Level 5 capability now or on any guaranteed timeline. Tesla has projected "full autonomy" for years, and while FSD has improved substantially, it remains a supervised system. The purchase is for advanced driver-assist, not a chauffeur.
Frequently Asked Questions
Is Tesla's Full Self-Driving Capability actually self-driving?
No. As of 2026, Tesla FSD is classified as SAE Level 2 — a driver-assistance system. The driver must keep hands on or near the wheel, monitor the road through the camera-based attention system, and be prepared to take over immediately. The "Full Self-Driving" name has been the subject of regulatory scrutiny and litigation. Tesla now appends "Supervised" to the feature name in most communications, acknowledging the human oversight requirement.
Where can I hail a driverless robotaxi right now?
Waymo One operates commercially in Phoenix (metro-wide), San Francisco (most of the city), Los Angeles (expanding zones), and Austin (limited). Zoox operates a pilot in Las Vegas. Cruise has a limited relaunch in Phoenix. Service boundaries are geofenced — you must be within the approved zone for both pickup and drop-off.
How much does a Waymo ride cost compared to Uber?
Waymo pricing is generally comparable to UberX or Lyft standard fares in the same market — roughly $2.40 to $3.60 per mile for a typical urban trip, with base fares around $5–$8. Unlike rideshare, there is no tipping. Surge pricing exists but tends to be less aggressive than Uber's peak surges. Waymo has introduced subscription plans in some markets offering discounted per-ride rates.
Are robotaxis safer than human drivers?
Available data suggests robotaxis have lower crash rates per million miles in their operational domains compared to human urban-driving benchmarks. Waymo reports roughly 76% fewer property-damage crashes and 85% fewer injury-involved crashes than human baselines. However, direct comparison is methodologically complex — robotaxis drive in mapped urban zones at moderate speeds, while human crash data includes rural highways, adverse weather, and impaired driving. Robotaxis are not immune to failures, as demonstrated by Cruise's 2023 San Francisco incident.
What happens if a robotaxi encounters a situation it cannot handle?
Level 4 robotaxis are programmed to execute a "minimal risk condition" — typically pulling over safely, activating hazard lights, and contacting a remote-assistance operator who can guide the vehicle or dispatch human help. The vehicle will not proceed through a situation it cannot confidently navigate. This is a key safety advantage over Level 2 systems, where the "fallback" is a human who may not be paying adequate attention.
Will Full Self-Driving ever become truly autonomous (Level 4 or 5)?
Tesla's stated goal is to achieve unsupervised full autonomy via software updates to existing hardware, but no timeline has been met, and regulatory approval for driverless Tesla operation has not been granted in any jurisdiction. Achieving Level 4 without lidar or geofencing remains an unsolved engineering challenge. Industry consensus among robotics and AI researchers is that camera-only, non-geofenced Level 5 autonomy is likely many years away — if it is achievable with current approaches at all. Purchase FSD for what it is today, not for what it might become.
Sources: NHTSA Automated Vehicle Safety, Insurance Institute for Highway Safety, Waymo Safety Reports, Tesla Vehicle Safety Reports, state DMV/CPUC public filings.
