Can self-driving cars fit into every city? Experts weigh in

Expert data confirms city limits on robotaxis

Experts agree that full Level 4 autonomy is not ready for every city. Analysts expect robotaxis to reach large, commercial-scale operations across roughly 40–80 cities worldwide by around 2035, rather than in every city simultaneously.

This limited deployment highlights the current constraints of the technology. Based on the type of power system, the market for self-driving vehicles is still primarily focused on semi-autonomous cars, which made up 95.19% of the total market in 2024.

Older streets demand 2025 tech improvements

Building on the problem of widespread deployment, older city infrastructure complicates the technology’s readiness. Experts from the University of New South Wales note that the current road safety rules were not designed with autonomous vehicles in mind.

Peer-reviewed studies show AVs are frequently struck from behind in urban settings, likely due to conservative behavior at intersections; exact percentages vary by dataset and city. This lack of predictable infrastructure increases the need for constant, complex software updates to ensure safe driving.

High hardware costs restrict private car sales

Since complex software requires expensive hardware, the high price of the self-driving technology is a significant barrier, according to market analysts. The cost of advanced components, such as Lidar sensors, makes full autonomy too expensive for most consumers.

ABI Research forecasts that nearly 36 million Lidar units will ship in 2025 for all vehicle types, showing the industry’s massive investment in expensive hardware. Due to these costs, only about 4% of new personal cars sold by 2035 are expected to feature Level 4 capabilities, says a World Economic Forum report.

Safety data shows need for human monitoring

Beyond the cost and technological challenges, the necessity of human oversight for safe scaling is a clear consensus among experts. Research published in an MDPI journal states that, until fully autonomous vehicles are introduced, even high-level self-driving cars occasionally require some level of human intervention.

NHTSA’s crash study assigns the ‘critical reason’ to human drivers in about 94% of cases, a statistic that does not equate to legal fault or sole causation. However, Level 4 systems still encounter “edge cases,” or complex scenarios, confirming the continued need for remote operators.

Robotaxi costs are dropping in core areas

Despite the challenges of human monitoring, the economic benefits are clear: the operational cost of robotaxi rides is projected to fall significantly, which may help cities become more viable. In Austin, Texas, a Tesla Robotaxi service launched in 2025 with an initial flat fare of $4.20 per trip.

Goldman Sachs Research estimates that the driving costs for commercial robotaxis will be $3.13 per mile in 2024, but may decrease to less than $1 per mile by 2030. This economic benefit incentivizes companies to expand in core urban zones where volume is high.

Tech leaders partner to deploy 5,000 Level 4 cars

The falling cost per mile accelerates partnerships between major car and technology companies, creating partnerships to scale up Level 4 cars in the U.S. and globally.

Stellantis and Uber announced plans, supported by NVIDIA’s DRIVE Hyperion 10 platform, to deploy up to 5,000 Level 4 (L4) vehicles in select cities, subject to testing and regulatory approvals.

The collaboration uses the NVIDIA DRIVE AGX Hyperion 10 platform, which includes the safety-certified DriveOS operating system. This strategic, combined effort aims to create safer, scalable, and more efficient mobility in defined operational zones.

Complex liability shifts from drivers to the car

As these automated vehicles hit the streets, experts note that introducing Level 4 vehicles fundamentally shifts liability in city accidents. When a fully autonomous car causes a crash, the responsibility is expected to move from the human driver to the automaker or the software provider.

Insurance companies must adapt, focusing on product liability insurance in addition to personal auto policies. In major regulatory areas, such as California, autonomous vehicle operators are required to hold a substantial insurance bond, typically $5 million, to operate commercially.

Weather conditions severely challenge city sensors

In addition to legal complexity, severe weather remains one of the most significant technical hurdles for widespread urban deployment, according to tech experts. Autonomous vehicles rely on sensors like LiDAR and radar, which are affected by heavy rain or snow.

Under severe rainfall, studies have measured mm-wave radar range reductions approaching ~45%, underscoring weather-hardening as a design priority. AV developers use wipers, heated elements, air ‘knives,’ and washer systems to keep cameras/LiDAR clear in rain, mud, and snow, now common in patents and prototypes.

Smart city tech is key to traffic flow in 2025

A partial solution to weather-related issues and complex traffic is linking cars to the city. Integrating self-driving vehicles with a city’s intelligent traffic systems is essential for improving congestion. Experts agree that Vehicle-to-Everything (V2X) communication is necessary for a seamless flow of traffic.

For example, a vehicle communicating with a traffic light can receive real-time data to optimize speed and avoid sudden stops. Cities like Beijing are rapidly expanding their autonomous driving demonstration zone to approximately 3,000 square kilometers to support this connected infrastructure.

The China advantage in fast urban scaling up

This focus on connected infrastructure helps explain why many experts point to China’s rapid and coordinated approach to urban deployment as a key development in 2025. Chinese cities have developed vast test zones with advanced infrastructure like smart traffic signals and 5G connectivity.

Baidu’s Apollo Go now operates one of the largest global robotaxi fleets, with over 1,000 fully driverless vehicles globally. This state-backed strategy enables companies to roll out services and gather crucial real-world data much faster than in the U.S.

Fewer parking spots could reclaim city space

Deploying these fleets brings a significant non-driving benefit of autonomous cars in cities: the potential to free up vast public space.

Researchers, such as those at Princeton University’s Alain L. Kornhauser, suggest that parking spaces will become obsolete in city centers because cars will no longer need to stay with their passengers.

Studies show that optimized autonomous vehicle parking lots could accommodate 62% more cars in the same area. This shift could reclaim valuable city land for pedestrians and public parks.

Want to know if the world is truly prepared for self-driving electric cars? Get the facts: Are we ready for Autonomous EVs?

Future city integration depends on shared vehicles

The goal of reclaiming city space and integrating self-driving cars seamlessly into every city depends heavily on the ownership model. Experts believe that if autonomous vehicles are shared instead of being privately owned, the need for parking could be reduced by as much as 90%.

Private vehicles are parked 95% of the time; models of shared autonomous fleets project far higher utilization (city- and scenario-dependent). This shared model is crucial for reducing congestion and fulfilling the promise of city transformation.

Curious how top brands stack up in self-driving tech? See the comparison in comparing autonomous-driving features across leading brands.

Could your city handle a driverless future? Please tell us what you think.

Read More From This Brand:

• How climate policies are pushing automakers toward EVs
• Volkswagen and Uber launch autonomous EV taxi service
• Tesla Debuts Fully Autonomous Cybercab Robotaxi In Paris

Don’t forget to follow us for more exclusive content right here on MSN

If you liked this story, you’ll LOVE our FREE emails. Join today and be the first to get stories like this one

This slideshow was made with AI assistance and human editing.