Estimating Situational Awareness and Predicting Gaze Allocation Strategies for Pedestrians Using a Markov Model

This study aims to measure and compare the situational awareness of pedestrians while crossing the road at unsignalized and signalized intersections in real-world conditions. The measurement of situational awareness employs Markov gaze entropy, and the obtained transition probability matrix is used to comprehend gaze transition behavior. This study conducted field experiments at an unsignalized and a signalized intersection with volunteer participants. Situational awareness of pedestrians was analyzed considering six areas of interest: Vehicles, Fellow Pedestrians, Near Path, Focus of Expansion, Road Infrastructure (RI), and Non-Traffic Relevant Objects (NTRO). Results of this study indicate that pedestrians at the signalized intersection exhibited lower situational awareness (i.e., higher entropy) than those at the unsignalized intersection. Additionally, higher crossing initiation times and pedestrian speed were associated with lower situational awareness at an unsignalized intersection. At the signalized intersection, pedestrians initiating crossing in 3 to 6 s exhibited least Markov Entropy, indicating high situational awareness. Furthermore, an increase in pedestrian speed was associated with increased situational awareness. Moreover, pedestrians at the unsignalized and signalized intersections exhibited more gaze transitions from NTRO to vehicles and from vehicles to NTRO. However, the highest average gaze transition probability (i.e., 49.6%) at the signalized intersection was observed between NTRO and vehicles, and the next highest probability (43.1%) was observed between RI and vehicles. Overall, these study findings can help gain insights into how pedestrians visually explore their surroundings and make decisions while crossing the road. This information can be valuable for designing safer and more efficient intersections, thus improving pedestrian safety.