Autonomous vehicles at Level 4 deployment enhance urban mobility

What's this about?

People disagree about whether Level 4 self-driving cars can make city travel better. These cars drive themselves only in set places and weather.

What supporters say

• Self-driving rides could help older people, disabled people, and others who cannot drive easily.
• They could give quick rides to bus stops and train stops.
• Shared self-driving cars may need fewer cars when many people need rides nearby.
• Fewer crashes could mean fewer injuries and fewer traffic jams after crashes.

What critics say

• Cities will not gain these benefits just by putting self-driving cars on streets.
• Self-driving rides may cost too much for people with low incomes.
• Companies could focus on busy areas instead of places with few transport choices.
• These cars could hurt public transport if they replace buses or trains.

The bottom line

Self-driving cars can improve city travel, but only with careful planning. Cities should keep buses and trains for busy routes and use shared, low-cost rides for gaps.

Level 4 autonomous vehicles — cars that can drive themselves within defined areas and conditions — can improve urban mobility, but only under the right conditions. The evidence does not show that simply putting these vehicles on city streets will reliably make transport more accessible, efficient or fair.

The case for

The clearest potential benefit is expanded travel for people who cannot easily drive themselves. That includes some older adults, disabled people and riders who depend on public transport. A federal evaluation of the Valley Metro–Waymo RideChoice program in the Phoenix area found that autonomous vehicles could offer on-demand trips for eligible riders and complement transit services. 1

Autonomous vehicles could also help solve the often difficult “first and last mile” of a journey: getting people from home to a rail station, bus stop or final destination. Pilot programs involving automated shuttles have pointed to possible benefits for local circulation and connections to transit. Modeling studies suggest that shared autonomous fleets can use fewer vehicles and provide better service when demand is dense enough and fleets are properly sized. 2

Safety is another possible advantage. Automation could reduce crashes, injuries and the disruption that follows them. Comparative real-world studies have found different crash patterns between automated vehicles and human-driven cars, while research on an established rider-only fleet reported lower rates for some types of crashes than human-driver benchmarks. 3

But these gains depend heavily on how cities deploy the technology. The strongest case is for vehicles that are shared, affordable, accessible and used to extend public transport networks rather than replace them. Public policy can steer autonomous services toward feeder routes and low-coverage areas while leaving high-capacity bus and rail lines to carry the main flow of passengers. 4

The case against

The central risk is that convenience for individual riders may create problems for the wider transport system. Autonomous vehicles could encourage people to make more trips, travel longer distances or choose cars instead of walking, cycling or taking transit. Vehicles may also drive empty while repositioning for their next passenger. All of that can raise total vehicle miles traveled and worsen congestion. 5

The effect on public transport is especially uncertain. Autonomous ride services could strengthen buses and rail if they bring riders to stations. But they could also draw passengers away from transit, particularly if they operate like cheap, taxi-style alternatives. The National Academies has concluded that both outcomes are plausible, and that coordination and public policy will largely determine which one occurs. 6

There are also practical limits. Current Level 4 services generally work only in restricted operating areas and under particular conditions. Studies of pilots and operating services have documented low speeds, limited routes, technical interruptions and cautious driving. Some services still require safety operators or other forms of human oversight. These constraints mean that successful pilots demonstrate local feasibility, not a broad transformation of city transport. 7

Accessibility, meanwhile, is not automatic. People with disabilities may need reliable boarding systems, wheelchair securement, clear communication, emergency help, suitable curb access and vehicles designed around their needs. Fares, data practices and uneven service coverage can also deepen inequities if governments and operators do not address them. 8

The bottom line

The evidence is balanced but conditional. Level 4 autonomous vehicles can enhance urban mobility when they are shared, pooled, affordable, fully accessible and integrated with public transit — and when rules limit empty driving and traffic growth.

But deployment by itself does not guarantee those outcomes. Poorly managed private or taxi-like services could add traffic, compete with transit and leave out riders who most need better options. Promising safety results apply to specific operating areas and fleets, not necessarily to every city or every Level 4 service.

Confidence in this conditional conclusion is high because different kinds of research point to both real opportunities and real risks. What remains missing is direct, citywide evidence showing sustained net improvements in access, congestion, transit performance, safety and equity. For now, the main lesson is clear: the technology’s impact depends less on its existence than on how cities choose to use it.