Investigation of pre-crash avoidance kinematics in pedestrians of different ages through volunteer experiment and scaling methodology

Objective: Understanding pedestrians' pre-crash avoidance kinematics is crucial for improving the identification of potential collision areas in interactions with highly automated vehicles (HAVs). Age significantly influences pedestrian avoidance velocity and the subsequent crash risks. However, current active safety systems in HAVs often overlook the influence of pedestrians' avoidance velocity and age on imminent accidents. This study analyzes how age affects pedestrian avoidance velocity and explores the incorporation of these factors in pre-crash scenarios to identify potential collision areas between pedestrians and vehicles.

Methods: Due to the infeasibility of measuring pedestrian avoidance behaviors in real-world pre-crash scenarios, we designed an indoor experimental platform replicating emergency crossroad scenarios to prompt subjects to mimic avoidance behaviors. 7 young and 7 middle-aged subjects participated in the experiment, resulting in a collection of 306 forward-avoidance, 297 backward-avoidance, and 42 normal-walking posture sequences. We developed a scaling approach integrating pedestrian kinematics and muscle physiology to establish a velocity-mapping relationship between young and middle-aged groups. Finally, we proposed an identification method for potential collision areas that considers pedestrians' age and avoidance velocity.

Results: Middle-aged subjects required more time for natural avoidance actions averaging 0.15 s for forward and 0.25 s for backward avoidance, compared to their younger counterparts. While the forward avoidance velocity of the middle-aged subjects exhibited an average decrease of 0.3 m/s compared to young subjects, their backward avoidance velocity remained nearly identical. Overall, middle-aged subjects have a larger potential collision area than young participants. Pedestrians who actively avoid vehicles have a smaller potential collision area compared to those who remain normal walking.

Conclusions: We developed an indoor simulated pre-crash scenario experiment and a scaling approach to reveal the correlation between pedestrian avoidance velocity and age. This method can be further applied to obtain the avoidance velocity of elderly pedestrians. Additionally, we validate the effect of these factors in assessing potential collision areas. The decrease in avoidance velocity highlights a larger potential collision area for middle-aged pedestrians when interacting with vehicles. Such facts and data shall be appropriately considered in developing intelligent protection systems for pedestrians.