NEW YORK, NY— New York City could soon face a new kind of traffic challenge: vehicles that can drive themselves but still need space on some of the most crowded streets in the country.
Robotaxis are already operating in cities across the United States, and companies are pushing to expand.
Now, Waymo, a subsidiary of Alphabet, has tested vehicles in New York City with a safety driver behind the wheel.
Before fully autonomous ride-hailing vehicles arrive on the bustling streets of the City, a new study from Hunter College tries to answer what could happen when thousands of driverless cars compete for road space, curb access and parking.
The report compares Waymo’s operating data with New York City taxi and ride-hail records.
Waymo vehicles have recorded fewer injury-causing crashes overall than New York City’s professional drivers.
But researchers found the company’s severe crash data remains limited and requires more study.
The analysis also found robotaxis could add traffic if vehicles spend more time traveling without passengers.
Waymo disputed the report’s findings, telling Streetsblog that its empty driving rate in California has been declining as the company improves fleet efficiency and vehicle use.
“The premise put forward by the report — that the New York City [taxi] fleet could fully turn over to Waymo vehicles — is also inaccurate,” Waymo spokesman Ethan Teicher told Streetsblog. “Waymo vehicles make up a fraction of the total ride-share vehicles on the road in the places where we operate, and that would also be true of New York City.”
Teicher added that Waymo’s fleet remains much smaller than New York City’s for-hire vehicle industry, noting that the company operates more than 3,500 vehicles nationwide, compared with more than 100,000 for-hire vehicle drivers in New York City alone.
The answer depends on the type of crash, the study finds.
Waymo vehicles recorded fewer crashes involving any reported injury than New York City’s for-hire vehicles.
Across all deployed locations, Waymo reported 0.71 injury-causing crashes per million miles driven, meaning its vehicles experienced fewer than one injury crash for every million miles traveled.
In the San Francisco Bay Area, where streets more closely resemble a dense urban environment like New York City, Waymo reported 0.70 injury-causing crashes per million miles.
By comparison, New York City’s for-hire vehicles recorded 2.15 injury-causing crashes per million miles in 2025.
That means Waymo vehicles recorded about 67 percent fewer injury-causing crashes per mile than New York City’s taxi and high-volume ride-hail fleet.
But the results changed when researchers examined crashes involving deaths or serious injuries, the measure used in Vision Zero safety analysis.
Waymo recorded 0.014 serious-injury or fatal crashes per million miles nationwide. The San Francisco Bay Area fleet recorded 0.030.
New York City’s for-hire fleet recorded 0.009.
Researchers found Waymo’s nationwide serious-crash rate was about one-and-a-half times higher than New York City’s for-hire fleet.
In the Bay Area, it was more than three times higher.
The report argues the comparison matters because robotaxis would replace professional drivers, not the average motorist.
"Robotaxis won't replace the average driver, they'll replace professional, for-hire drivers who historically have better safety records than the general public in New York City and beyond," the report states.
"The best comparison isn't robotaxis versus everyone, it's robotaxis versus the for-hire fleet they would replace."
Researchers said the number of severe crashes remains too small to determine long-term risk.
Waymo’s fleet had driven about 220.6 million miles when researchers analyzed the data. New York City’s for-hire fleet traveled more than 1.75 billion miles in 2025.
The difference affects confidence in the results.
New York City’s crash rate is based on a much larger number of miles, creating a narrower statistical range.
"Waymo's fleet has not yet driven enough miles to support that they are safer than FHV drivers," the report states.
The report identifies empty driving as one of the biggest challenges.
A human driver can stop working when demand falls. A robotaxi can continue operating as long as it has power.
Researchers analyzed Waymo’s California operating data to measure how often vehicles traveled without passengers.
In March 2026, Waymo vehicles spent:
Together, waiting and pickup travel accounted for 41.6 percent of miles driven without a passenger.
New York City’s high-volume for-hire vehicles recorded a 33 percent empty-driving rate.
Researchers calculated that Waymo traveled 1.71 miles for every passenger mile served, compared with 1.49 miles for New York City’s Uber and Lyft fleet.
The difference means replacing a human-driven ride-hail trip with a robotaxi could increase vehicle miles traveled by about 14.8 percent.
The curb could become one of the biggest challenges for autonomous vehicles.
Waymo does not publicly report how long vehicles spend parked while waiting for passengers.
Researchers estimated parking behavior using publicly available operating data.
They estimated Waymo vehicles spent 57.7 percent of their waiting time parked in California.
That equals about 10.4 minutes per trip, or roughly four hours of parking time per vehicle each day.
In New York City, where curb space is limited, researchers said that behavior could create new pressure on already crowded streets.
Researchers modeled what would happen if robotaxis had fewer places to wait.
They found every 1 percent decrease in parked time could increase empty driving by about 1.5 percent.
If robotaxis could park only 33 percent of the time while waiting, compared with the estimated California rate of 57.7 percent, the report projected deadheading would increase by 37 percent.
That could mean more vehicles circulating without passengers.
The report also raised questions about where robotaxi companies would store, charge, clean and maintain vehicles.
If depots were located outside Manhattan, such as in New Jersey or the outer boroughs, additional empty trips to and from those facilities could add traffic on bridges and tunnels.
Robotaxi companies often point to one advantage: fleets of autonomous vehicles can share information.
A vehicle could learn about a blocked road, emergency or traffic condition and adjust routes.
But researchers said that benefit depends on how cities regulate companies.
If one company operates the entire fleet, vehicles could coordinate more easily.
If multiple companies operate separately without sharing information, each fleet may compete for the same limited curb and road space.
The report warns that without data-sharing rules, cities could end up with separate fleets that do not communicate with one another.
The report recommends requiring robotaxi companies to compare their safety records with New York City’s professional taxi and ride-hail drivers before receiving operating approval.
Researchers argue that professional drivers should serve as the benchmark because they are the group robotaxis would replace.
They also recommend independent review of safety studies.
Researchers said cities need more information to understand how autonomous fleets affect neighborhoods.
They recommend requiring companies to report:
Researchers said cities should continue receiving that information after approval to monitor long-term effects.
The report also recommends considering fees tied to how much space robotaxis use.
Possible approaches include:
Researchers said pricing could encourage companies to reduce unnecessary driving.
The report raises a broader question: Should autonomous technology focus only on private ride services?
Researchers point to automated shuttle programs already operating in some locations and argue cities could use automation to fill transit gaps.
They recommend examining whether autonomous technology could expand public transportation rather than only increase private vehicle trips.
Researchers recommend studying limits on robotaxi access in dense commercial districts.
Areas such as Midtown Manhattan, the Financial District and Downtown Brooklyn already face heavy congestion.
A policy known as geofencing could restrict robotaxis from entering certain areas or direct them toward neighborhoods with fewer transit options.
Robotaxis could add pressure to the city’s crowded curb system.
Researchers recommend expanding designated pickup and drop-off zones for ride-hail vehicles, whether or not robotaxis arrive.
The city would need to decide:
Because robotaxis do not have human work-hour limits, researchers said fleets could operate for longer periods than current ride-hail vehicles.
The report recommends monitoring whether robotaxis increase total vehicle miles traveled and considering limits on fleet size if they add congestion.
Sign up for free local newsletters and alerts for the
New York City, NY Patch
Patch.com is the nationwide leader in hyperlocal news.
Visit Patch.com to find your town today.