Table of Contents
A University of Wisconsin-Milwaukee doctoral researcher is combining artificial intelligence, computer vision and transportation engineering to investigate how lighting, driver behavior and intelligent vehicle systems could help prevent pedestrian crashes.
MILWAUKEE: The seconds before a road accident can determine whether a pedestrian escapes danger or suffers a potentially fatal collision. A researcher at the University of Wisconsin-Milwaukee (UWM) is studying how artificial intelligence and advanced video analysis can help drivers and automated vehicles recognize pedestrians earlier.
Muhammad Fahad, a doctoral researcher at UWM, is working at the intersection of transportation engineering, computer vision and artificial intelligence. His research examines how drivers perceive pedestrians, how dangerous interactions develop and how intelligent transportation systems could identify risks more effectively.
Nighttime Crosswalk Research in Wisconsin
Fahad is contributing to an enhanced crosswalk illumination study supported by the Wisconsin Department of Transportation (WisDOT). Led by Professor Xiaowei (Tom) Shi, the project examines pedestrian visibility at two locations: a four-lane road in Madison and a two-lane road in Milwaukee.
Researchers are testing seven lighting spectra, along with different fixture positions and mounting heights, to determine how lighting conditions influence a driver’s ability to recognize a pedestrian.
During nighttime experiments, a mannequin is placed near a crosswalk as an instrumented vehicle approaches. Cameras record the roadway and driver, while positioning equipment, vehicle-speed measurements and braking data help researchers assess when the pedestrian becomes recognizable.
Fahad’s work includes analyzing video footage and connecting pedestrian-recognition events with vehicle movement and braking responses.
The objective is to determine whether particular lighting arrangements improve detection distance and give drivers more time to react. Simply making a crossing appear brighter does not necessarily mean a pedestrian will be recognized sooner.
At 40 miles per hour, a vehicle travels nearly 59 feet every second. Even a brief improvement in recognition time could therefore be important, although the effectiveness of individual lighting designs must be established through controlled testing and further validation.
AI Helps Researchers Understand Road Conflicts
Fahad is also involved in CrossTraj, a research effort that uses roadside camera footage to reconstruct pedestrian and vehicle movements around urban crossings.
By examining movement patterns, researchers can study situations involving hesitation, yielding, sudden changes in direction and potentially dangerous approaches.
Unlike crash records, which document incidents after they occur, trajectory analysis can help explain what happened in the moments leading up to a possible collision. The findings may also help researchers evaluate whether autonomous vehicles can interpret unpredictable pedestrian behavior.
Self-driving systems must recognize people who may pause, step into the road or suddenly change direction. Real-world movement data can help test the ability of computer-vision and prediction models to handle such situations.
Another area of Fahad’s research focuses on nighttime headlight glare and its effects on human drivers and camera-based perception systems.
Why Nighttime Pedestrian Safety Matters
The research addresses a significant road-safety challenge in the United States.
According to a preliminary analysis released by the Governors Highway Safety Association in July 2025, an estimated 7,148 pedestrians were killed on U.S. roads in 2024. More than three-quarters of pedestrian fatalities occurred after dark, highlighting the importance of nighttime visibility.
The Federal Highway Administration has also identified roadway lighting as an important pedestrian-safety measure. Its research has reported reductions in nighttime pedestrian-injury crashes associated with intersection lighting, although those broader findings should not be interpreted as results already achieved by Fahad’s specific project.
Academic Research and Leadership
Fahad began his academic journey in mechanical engineering at Tsinghua University in Beijing before pursuing research in transportation safety, intelligent vehicles and computer vision at UWM.
His published work includes studies in engineering journals such as IEEE Transactions on Intelligent Transportation Systems and IEEE Access. He has also reviewed research manuscripts for engineering publications.
At UWM, he contributes to autonomous transportation research associated with the U.S. Department of Transportation-supported TRAVELS Center, which examines autonomous vehicles and transportation challenges in rural environments.
Beyond research, Fahad serves as president of the Pakistan Student Association at UWM. He previously held the same position at Tsinghua University and has received academic recognition, including a Young Research Scientist Award and Student of the Year recognition associated with his earlier studies.
The Goal: Recognize Danger Before Impact
Fahad’s research brings together roadway lighting, pedestrian movement analysis and intelligent vehicle perception to investigate a shared problem: how to identify danger before a collision occurs.
Although no single technology can eliminate road accidents, better evidence about visibility, driver responses and pedestrian behavior could help transportation agencies improve crossings and guide the development of safer automated vehicles.
Ultimately, understanding what happens in the seconds before impact may help researchers develop practical solutions to protect pedestrians on roads across the United States.
