Missouri recorded a historic high for pedestrian deaths in 2024 — 148 — a 16% increase from 2023 and the highest ever reported in the state, according to the Missouri Department of Transportation. A new research project at the University of Missouri is looking to make streets safer for everyone.
Harnessing the power of light detection and artificial intelligence, researchers at MU’s College of Engineering hope to eliminate traffic fatalities among vulnerable road users like pedestrians and bicyclists. The initiative is part of the federal “Vision Zero” program.
“We’ve seen increasing fatalities involving pedestrians and cyclists who share the road with cars,” associate professor Yaw Adu-Gyamfi said. “This project helps autonomous and traditional vehicles better understand human behavior in traffic scenarios.”
Adu-Gyamfi is leading the project alongside graduate student Linlin Zhang.
“This isn’t just about futuristic technology,” Adu-Gyamfi said. “It’s about using data to save lives today.”
He notes that in Columbia, poor nighttime lighting and limited visibility at intersections increase the risk for pedestrians. Traditional traffic monitoring systems, which rely primarily on cameras, often fail to accurately measure speed and distance — particularly in low-light or rainy conditions.
In contrast, lidar sensors generate reliable spatial data by producing a three-dimensional view of the environment. By combining lidar data with video footage, the research team is training models to recognize human behaviors, such as a pedestrian preparing to cross the street or a cyclist merging into traffic.
In addition to identifying pedestrian behavior, the system can detect “near-miss” events by tracking sudden changes in speed or direction. These incidents, where a crash is narrowly avoided, serve as early indicators of intersection-level risk.
“If we observe a high number of ‘close calls’ at a particular location, it often predicts a future crash,” Adu-Gyamfi said.
Another key issue Adu-Gyamfi’s team identified is the inadequacy of current traffic signal timing.
“In many places — including here in Columbia — the green light duration is too short for people who walk slowly, like elderly residents or individuals with mobility challenges,” Adu-Gyamfi said. “Young people may be able to make it across in time, but many others cannot. That’s a safety risk we can fix.”
According to the Missouri Department of Transportation, the average walk interval is typically around seven seconds, though it can vary, with a minimum of four seconds or more depending on the intersection’s needs and engineering judgment.
Integrating lidar with traffic signal systems could allow real-time adjustments based on pedestrian flow. For example, if sensors detect an elderly person crossing, the system could extend the green light automatically.
The team plans to expand testing to St. Louis in July, where city officials have approved installation of equipment at eight intersections. The St. Louis phase will provide a broader dataset that includes elderly individuals and people with disabilities, helping the researchers build more inclusive behavioral models.
Lidar generates a massive amount of data — up to five gigabytes per minute. To manage this, the researchers are developing real-time processing systems that eliminate the need for long-term storage.
High-performance computers used for data processing can also be expensive. However, Adu-Gyamfi said multiple intersections connected to the same traffic controller can share one computer, making the system more affordable and scalable.
“The ultimate goal is not just to predict behavior but to act on it — giving people enough time to cross safely,” Adu-Gyamfi said. “We want to make sure our streets are designed for everyone, not just the fast and able-bodied.”
