Eyes on the Road

Article Content

This story is from the 2026 issue of Discoveries, a UC San Diego Health Sciences magazine.

Bright orange trailers line the highway, looking nothing like typical roadside decorations, and with no signs or labels, they seem to be silently watching the flow of cars.

In a sense, they are.

They are part of a research effort led by the Transportation Research and Education for Driving Safety (TREDS) Center at UC San Diego designed to better understand how people actually behave behind the wheel. All around San Diego County, the trailers were deployed with radar and infrared cameras to capture commercial motor vehicle (CMV) driver patterns to better assess the prevalence of high-risk behaviors.

Artificial intelligence (AI)-powered cameras captured round-the-clock data, revealing that 11% of CMV drivers engaged in risky driving behaviors, including speeding, handheld phone use or failure to wear a seatbelt — despite California laws and fines targeting these behaviors.

Preventable human behaviors such as these are primary causes of collisions, accounting for up to 95% of motor vehicle traffic fatalities.

“These findings highlight how commonplace risky behaviors are on our shared roadways. Leveraging AI allows for a clearer understanding of behaviors in real-world contexts, which can enable informed and targeted interventions to save lives,” said Ryan Moran, MD, MPH, co-director of TREDS at the UC San Diego Herbert Wertheim School of Public Health and Human Longevity Science.

TREDS focuses its research and programming on preventing crash-related injuries and death by improving safe behaviors on the roadways.

Supplied by Acusensus, the cameras gathered data over 7-day periods at 16 sites across San Diego County, spanning rural and urban interstates and roads. In total, AI was used to process 2,600 hours of driving data.

TREDS received images with license plates and faces blurred for privacy, yet the high-definition images clearly captured drivers’ hands holding phones and seatbelt usage.

The analysis showed that speeding was the most common violation, followed by unbelted driving and handheld phone use, especially during early morning hours and weekends.

The team tested the feasibility of sending driver-specific but anonymous messaging in real time, giving drivers feedback about their own behaviors as they passed by a roadside message board. TREDS was able to solve logistical and technological challenges in setting up a functional system and plans to extend its program to additional Southern California counties.

“Continuing to refine data-driven approaches to traffic safety and collaborating with policymakers, transportation agencies, industry stakeholders and the public will be essential in leveraging these insights to create safer roadways for all,” said Jill Rybar, MPH, TREDS associate director. “Every driver shares the responsibility to operate safely and protect our community.”

Driven by data

Today, TREDS operates as a vital center with 18 faculty and staff, but its origins trace back to a single project. In 2007, Linda Hill, MD, MPH, TREDS director and Distinguished Professor at the Herbert Wertheim School of Public Health, received a California Office of Traffic Safety grant to work within hospitals to provide driving safety screening and management to adults over 65 and develop clinical screening guidelines to help doctors identify age-related driving impairments.

In that year, the United States recorded 43,945 traffic fatalities. The toll declined to a low of 35,303 in 2015 before beginning to rise again.

In 2023, motor vehicle crashes claimed 44,762 lives nationwide, with 4,061 deaths in California alone.

“The majority of roadway collisions are not random accidents. Car crashes and the resulting deaths are not inevitable acts of nature. They represent a public health threat we have the power to mitigate by changing our actions behind the wheel,” said Hill.

To address the public health dangers of high-risk driving behaviors, the TREDS team leverages expertise in public health, medical engineering, law enforcement and neuroscience to create educational driving safety programs tailored to both general and specific populations, including teenagers, seniors, Indigenous populations and CMV drivers.

In 2025, TREDS convened a statewide summit that gathered California’s commercial driver industry to develop strategies for reducing impaired driving.

Road safety, reimagined

Although education is part of TREDS’s core mission, recent grants have enabled technology-driven research to understand the behaviors that lead to collisions and devise novel interventions that help drivers adopt safer habits.

These projects focus primarily on commercial drivers, who represent a significant share of roadway users in California. The state has 691,094 registered CMVs.

CMVs pose distinct hazards for drivers and passengers. From 2020 to 2022, nationwide fatalities involving large trucks rose 20% from 4,945 to 5,936.

“With the momentum and size of commercial vehicles, crash outcomes are more dire, both for the commercial vehicle driver and for other road users,” said Moran, who also serves as a professor in the UC San Diego School of Medicine and the Herbert Wertheim School of Public Health. “It also puts an enormous toll on the road network because a crash with a commercial motor vehicle is a much harder thing to clear up.”

Geofencing in action

Although large trucks comprise only 5% of total vehicular traffic within work zones, they were involved in 30% of fatal work zone crashes nationwide in 2022.

To improve safety, TREDS recruited 382 commercial fleets for a study investigating the impact of hands-free, real-time safety alerts delivered through in-cab electronic logging devices (ELDs). Anonymous, second-by-second data from 228,000 CMVs traversing 4,080 work zones were captured for 30 seconds before and up to 5 minutes after an alert was displayed.

The experimental cohort received traffic alerts via their Drivewyze ELD once they entered a geofence, indicating they were approaching a Caltrans work zone; the control group received no alerts.

Results showed that drivers who received alerts about delays, detours, slowing traffic or lane closures slowed down significantly more than drivers who did not receive alerts. A follow-up survey indicated strong driver support, with 82% reporting slowing down and 83% saying they paid closer attention after the alerts.

“Work zones are among the riskiest environments for truck crashes,” said Hill. “Our research shows that real-time alerts encouraging safer and more attentive driving are not only effective at changing behavior but are also welcomed by drivers themselves.”

Read more news about: The Herbert Wertheim School of Public Health and Human Longevity Science, Health Sciences