Skip to content
The Future Explorerby Benjamin Talin
Enablertechnological

Autonomous Vehicles

Cars, buses and lorries that drive themselves, at first inside carefully mapped areas, without a human at the wheel. Cameras, radar and lasers watch the road while software decides what to do. A famous prediction once put a million robotaxis on the road within about a year; there were none. Today paid driverless rides run in several US cities, hundreds of thousands a week. The driving works there; laws, insurance and trust are catching up.

Maturity

3/ 5
measured
BasisDriving capability proven in defined areas, with paid driverless service at scale in several US cities. Liability, approval and acceptance lag well behind.SourceCPUC approvals (10 Aug 2023; 1 Mar 2024); Abdelhalim, Findings (18 May 2026); Automated Vehicles Act 2024 (c. 10); California DMV (24 Oct 2023)Vintage2023–2026
ReadinessCapability 4Deployment 3Absorption 4Bound byregulationTrendimproving

Where it feeds in

4 megatrends in 2 forces

Where you have already met this

In 2019 a confident, dated prediction put a million robotaxis on the road by the end of 2020. There were none. Paid driverless services now run in several cities - years later, inside mapped zones, and with nobody in the driver's seat in the leading service.

If this moved

The prize is not transport, it is land. A large share of urban space is given to parking for vehicles idle 95% of the time. Changing that assumption changes the physical form of the city, which is why this sits under the built-environment megatrend as well as mobility.

What to watch for

Liability allocation, which is the actual constraint. Type approval that generalises beyond a single city. Removal of safety monitors, which is the honest marker of readiness.

Not to be confused with

Driver assistance. The gap between a system that helps and one that is responsible is the entire regulatory problem, and marketing language routinely blurs it.

What it is

Road vehicles capable of driving without human intervention, in defined conditions or generally.

How it works

  1. 01Sensors build a pictureLidar, radar and cameras map everything around the vehicle, many times a second.
  2. 02The car finds itself on a mapLive sensor data is matched against a detailed, pre-surveyed map to fix the car's position to centimetres.
  3. 03Software predicts what others will doModels classify cars, cyclists and pedestrians and estimate where each will be in the next few seconds.
  4. 04A planner picks a pathThe system chooses a safe route that follows the traffic rules.
  5. 05The car executes it - inside its zoneSteering, throttle and brakes are controlled electronically. All of this is approved only for a defined area and set of conditions.

State of play

  • Waymo gave more than 100,000 paid rides a week in August 2024 (Waymo co-CEO, 20 Aug 2024, via TechCrunch) and over 150,000 by October 2024 (Alphabet Q3 2024 earnings call, 29 Oct 2024).
  • Waymo reported 500,000 paid rides a week across 10 US cities in March 2026 (Waymo, via TechCrunch, 27 Mar 2026).
  • California robotaxis made 13,790,147 trips between August 2023 and December 2025; 53.6% of their miles carried a passenger (Abdelhalim, Findings, 18 May 2026, from CPUC data).
  • California's DMV suspended Cruise's driverless permits on 24 October 2023, finding its vehicles not safe for public operation and that the company had misrepresented safety information (California DMV, 24 Oct 2023).
  • The UK Automated Vehicles Act 2024, which received Royal Assent on 20 May 2024, makes the authorised self-driving entity responsible for the vehicle and protects the user-in-charge from driving offences while the system drives (Automated Vehicles Act 2024, c. 10).
  • US federal safety exemptions remain capped at 2,500 vehicles per manufacturer per year (49 U.S.C. §30113; NHTSA, 13 Jun 2025).

Why it matters

It would restructure how cities are built and how much of urban land is given to parking and roads. The technical problem has been substantially solved in defined conditions, and the ramp is now bound elsewhere.

  • The driving works inside defined zones. What holds the mobility megatrend at 2 is regulatory clarity, not software.
  • It is the book's cleanest failed forecast: a million robotaxis promised for 2020, none delivered, then a real service arriving years later by a different route.
  • The prize is land, not transport. A city that needs less parking can be built differently.

What holds it back

Liability and approval

Who is responsible when the car drives is being settled law by law. The UK wrote it into statute in 2024.

Geography

Service works inside mapped, permitted zones. Every new city is a new approval and a new map.

Trust after incidents

One incident grounded a whole fleet in 2023. Safety records are built slowly and lost quickly.

Terms

Operational design domain (ODD)
The area and conditions - roads, weather, speed, time of day - a driverless system is approved to operate in.
SAE Level 4
Driving automation that needs no human driver, but only inside its operational design domain.
Sensor fusion
Combining lidar, radar and camera data into one picture of the surroundings.
HD map
A pre-surveyed map accurate to centimetres, with lanes, kerbs and signals.
Disengagement
A moment when the automated system hands control to a human or a remote operator.

Steepens

MegatrendWeightRole
Autonomous Systems and Mobilitylarge
The Built-Environment Deficitmoderate
Automation of Cognitive Workmarginalmarginal: not counted as a weakest link
Energy Transition and Electrificationmoderate

The gradient behaves like a sticky rubber band: it is pulled hardest toward the weakest enabler. Marginal links are shown but do not pull.

If this moved one step

Computed from the map as it stands: this score raised from 3 to 4, every other score unchanged. It is arithmetic, not a forecast.

Raised one step on its own, it lifts no megatrend's weakest link; each gradient it feeds still rises a little.

The gradient also rises in

What if it moved?

What would change the gradient

Conditional, not predictive. No dates, no probabilities.

  1. 01Liability allocation - the actual binding constraint
  2. 02Type-approval regimes that generalise beyond a single city
  3. 03Public acceptance following visible safety records

Common misconception

That the remaining problem is technical. Commercial robotaxi services operate today; liability frameworks, insurance, approval regimes and public acceptance are each less mature than the driving.

Talin, B. (2026). Autonomous Vehicles. The Future Explorer, dataset v0.4.0 (2026-11-16). https://explorer.talin.digital/e/autonomous-vehicles/

Autonomous Vehicles: maturity 3/5 · The Future Explorer