Research of fatal car-to-pedestrian precrash scenarios for the testing of the active safety system in China. (February 2021)

• Record Type: Journal Article
• Authors: Tan, Zhengping; Che, Yaoyue; Xiao, Lingyun; Hu, Wenhao; Li, Pingfei; Xu, Jin
• Is Part Of: Accident analysis and prevention. Volume 150(2021)
• Publication Date: 2021-02
• Subjects: Traffic accident -- Pedestrian -- Test scenario -- Autonomous driving -- Automated emergency braking (AEB) -- Cluster analysis; Accidents -- Prevention -- Periodicals; Accident Prevention -- Periodicals; Accidents -- Prévention -- Périodiques; 363.106
• DOI: 10.1016/j.aap.2020.105857
• Languages: English
• ISSNs: 0001-4575

Abstract

Highlights: A novel method was proposed to define the basic scenario by extracting crucial variables from description of precrash. A two-step implementation consisting of descriptive statistics and cluster analysis is used to mine detailed scenarios. A total of 34 basic scenarios and 13 detailed scenarios were obtained from fatal pedestrian accidents in China. Similarities and differences between fatal pedestrian scenarios in China and NCAP were compared and analyzed. Abstract: Road safety remains a challenge with numerous Vulnerable Road Users (VRUs) suffering from injuries and death every year. Pedestrian protection using active safety systems, such as Automated Emergency Braking (AEB), is an effective measure to combat the situation. Furthermore, the perception of precrash scenarios plays an important role in active safety research. It is essential to understand and define precrash scenarios. This study aimed to apply the obtained typical car-to-pedestrian precrash scenarios from Chinese severely injured pedestrian traffic accidents to develop and test active safety systems. The National Automobile Accident In-Depth Investigation System (NAIS) recorded 467 cases from 2011 to 2018 in China, and 12 items were selected from the NAIS database as description variables for the precrash scenario. The items were divided into four categories: car, pedestrian, road, and environment. Group decision theory was applied to evaluate the importance of each variable in its category. A total of 34 basic scenarios were defined and obtained according to the extracted significant variables. These basic scenarios represented diverse fatal scenarios in China which are crucial for autonomous driving. The frequency distribution of the scenarios demonstrated that the top five scenarios covered 85.3 % of the total. Five scenarios were identified to have the common characteristic of cars going straight. Additionally, 13 detailed scenarios were obtained from the five basic scenarios by using cluster and frequency analyses. In contrast to the New Car Assessment Program (NCAP) test scenarios, weather and lighting conditions were considered in these 13 scenarios, and the driving speed before the crash were mostly distributed in the range of 40–80 km/h (20–60 km/h in the NCAP). Meanwhile, both walking and running were commonly recorded for pedestrians to cross the street from the nearside, compared with records of walking only to cross from the nearside in the NCAP. These results contribute to a reference for test scenarios of pedestrian AEB or Forward Collision Warning (FCW) in China.