What Are Pedestrians?

Pedestrians are people traveling on foot, or using a personal mobility aid such as a wheelchair, within a road or transit environment. In traffic engineering and human factors research they are treated as a distinct road user class with their own movement dynamics, sensory limits, and injury tolerances, separate from drivers, cyclists, and vehicle occupants. The technical study of pedestrians draws on behavioral science, impact biomechanics, infrastructure design, and machine perception, the last of which has grown in importance because automated vehicles must detect and predict people on foot before they can operate in mixed traffic.

Engineering attention to pedestrians is driven by exposure and vulnerability. A person walking carries no crash structure and no restraint system, so the probability of a fatal outcome rises steeply with vehicle impact speed. Crash statistics compiled by NHTSA in its pedestrian safety review show that most pedestrian deaths in the United States occur in darkness and away from intersections, a pattern that points directly at lighting, crossing spacing, and speed management as design levers.

Pedestrian Behavior and Movement Modeling

Pedestrian flow is modeled at several scales. Microscopic models represent each walker as an agent subject to attraction toward a goal and repulsion from obstacles and other walkers, an approach popularized by the social force formulation of the mid-1990s and still used in evacuation simulation. Macroscopic models instead describe a crowd as a continuum and use a fundamental diagram relating pedestrian density to walking speed and flow rate, which underpins the level-of-service grades applied to sidewalks, stairs, and platform edges. At crossings, behavior is often described through gap acceptance: a walker estimates the time until an approaching vehicle arrives and steps out when that gap exceeds a personal threshold. Individual variation in walking speed, typically between about 1.0 and 1.5 meters per second for unimpaired adults, matters enough that signal timing standards specify a design walking speed rather than an average.

Detection and Intent Prediction

Automated pedestrian detection began with handcrafted image features and sliding-window classifiers and moved to convolutional and transformer architectures trained on annotated video. Benchmarks built from vehicle-mounted cameras exposed the hard cases that dominate real risk: small figures at long range, heavy occlusion, and nighttime scenes. Production systems now fuse visible-light cameras with radar, lidar, and sometimes thermal imaging so that detection does not collapse when illumination fails. Prediction is the harder half of the problem, since a safe maneuver depends on whether a person standing at a curb intends to cross. Intent is inferred from body pose, head orientation, and trajectory history. Validating that inference is difficult because crashes are rare, so evaluation leans on conflicts captured by roadside video systems, and studies of automatically generated pedestrian conflict surrogates test how reliably those detected events correspond to conflicts confirmed by review.

Infrastructure and Countermeasures

Roadway design changes remain the most reliable way to reduce pedestrian injury, because they act on the whole traffic stream rather than on individual attention. The FHWA Proven Safety Countermeasures set includes raised crossing islands, road diets that reconfigure four-lane roads, leading pedestrian intervals that give walkers a head start at signals, rectangular rapid-flashing beacons, and improved crosswalk lighting. Each is tied to a measured crash reduction factor. Vehicle-side measures complement these, including hood and bumper geometry rules that reduce head and leg injury, and automatic emergency braking calibrated for pedestrian targets. Severity analyses of crash records, including machine learning studies of pedestrian injury outcomes, rank lighting condition, road type, and weather among the strongest predictors of how severely a struck pedestrian is injured.

Applications

Pedestrian research has applications in a wide range of fields, including:

• Roadway and intersection design in traffic engineering
• Automated driving and advanced driver assistance systems
• Building egress modeling and evacuation planning
• Injury epidemiology and public health surveillance
• Urban planning and walkability assessment
• Crowd management at stadiums, transit hubs, and large events