52
People killed in Rhode Island traffic crashes in 2024, per NHTSA’s FARS comparison table. The 2025 early estimate is 37.
Independent · Ocean State
What the data say about autonomous vehicles in Rhode Island
RI for AV is a grassroots group. We think the published safety record of driverless ride-hail, and Rhode Island’s own crash and access numbers, are worth a serious look. We are not a company and not the state.
Numbers to start with
Every figure below is sourced. Operator-produced numbers are labeled as such.
271M
Waymo rider-only miles through June 2026 (no human in the driver’s seat), published by the company, mostly in five metro areas.
9.1%
Of occupied Rhode Island housing units had no vehicle available (ACS 5-year 2024; 44,362 units, ±0.4 percentage points).
0
Enacted Rhode Island statutes authorizing fully driverless vehicles. IIHS lists the state’s AV law as “none.”
How to read this site
Crash rates (events per mile) matter more than crash counts. AV companies must report even minor contacts to NHTSA; many human fender-benders never enter police files. Rhode Island winters are not the cities in the Waymo studies. We keep those limits next to the numbers, not in a footnote you have to hunt for.
Rhode Island still loses people on the road
NHTSA counted 52 traffic deaths in Rhode Island in 2024, a rate of 0.68 per 100 million vehicle miles. That is lower than the U.S. rate, and it is still 52 families. RIDOT’s separate crash repository counted 45 fatalities that year and 42 in 2025. The two systems do not match exactly; both are official, and both show a problem that has not gone away.
In 2025 RIDOT recorded 36,850 police-reported crashes, 268 serious injuries, 10 pedestrian deaths, and 9 motorcyclist deaths. Through September 30, 2026, the state repository already shows 36 fatalities, including 10 motorcyclists and 9 pedestrians.
RIDOT Highway Safety Improvement Program 2025 annual report, 2016–2024. 2017 and 2023 were high years. The series is not a straight decline.
What the AV evidence shows, and what it does not
Peer-reviewed rate comparison
Kusano et al. (2025), in Traffic Injury Prevention, compared Waymo rider-only driving through January 2025 (56.7 million miles in Phoenix, San Francisco, Los Angeles, and Austin) with human benchmarks built from state crash and mileage data. They found statistically lower rates for injury-reported, airbag-deployment, and suspected-serious crashes, including a 96% reduction in injury-reported vehicle-to-vehicle intersection crashes (87–99% CI). Authors are Waymo-affiliated. Events come from NHTSA’s crash order.
Insurance claims (operator-involved)
A Waymo analysis using Swiss Re liability data over 25.3 million fully driverless miles reported 9 property-damage claims and 2 bodily-injury claims, versus 78 and 26 expected for the human benchmark. That is an 88% and 92% reduction versus all human drivers, and 86% / 90% versus 2018–2021 cars with driver-assistance systems. Swiss Re is independent; Waymo is still in the study.
Methods, city miles, and the NHTSA reporting caveat →
Access, not just novelty
Census ACS 5-year 2024: 9.1% of occupied Rhode Island housing units (44,362) had no vehicle. RIPTA carried 13.1 million trips in FY 2024, including 288,736 on RIde paratransit for seniors and people with disabilities. Late buses, tourism in Newport and Providence, and people who cannot or should not drive are the local case for robotaxis, if the vehicles can handle our weather and if the rules protect riders and workers. We do not have a Rhode Island robotaxi employment or tourism study. We do not invent one.
Vehicle access, crash costs, transit →
Law and local action
The General Assembly has not passed an AV deployment law. It has seen an autonomous-vehicle bill (S 2514, 2016) and several personal-delivery-device bills (sidewalk robots). RIDOT ran Little Roady, a slow shuttle with a human attendant, on Providence streets in 2019–2020. Neighboring Massachusetts and New York, in the IIHS tracker, allow testing with a licensed driver in the vehicle. Connecticut, like Rhode Island, is listed as “none.”