VR-based Pedestrian Training Dataset (2018, Wuhan)

Objective: China has a high fatality rate for child pedestrians, which highlights the necessity of implementing more effective pedestrian safety training programs in elementary schools. We thus investigated the efficacy of using virtual reality (VR) as an instructional technology to identify and modify risky pedestrian behaviors among Chinese children. Methods: Seventy-nine children (grades 1 through 3) from three elementary schools in Hubei province participated and were categorized into urban (n = 20), migrant (n = 29), or rural (n = 30) students based on the schools’ locations. They completed a VR program comprising three street-crossing challenges to measure five pedestrian behaviors. The participants first attempted to complete the challenges by themselves in the first-time trial (T1) and then engaged in a personalized debriefing session before undertaking the challenges a second time (T2). Pedestrian performance for the two trials was compared by school location and grade level as between-subjects factors, and the rationale behind risky pedestrian behaviors was inductively analyzed. Results: Three risky pedestrian behaviors were observed in the program: dashing into the street, crossing on a blinking green light, and failing to check for traffic. Potential reasons for these behaviors included a lack of knowledge of road signs and traffic rules and the absence of daily adult supervision. The overall pedestrian performance increased from T1 to T2 with a moderate effect size (Ƞp2 = 0.59, P < .001). A significant main effect of the trials was found for the three pedestrian behaviors (for all values, P < .001); however, interactions of trial by location and trial by grade were nonsignificant in all univariate tests (for all values, P ≥ .05). Conclusions: VR is an effective technology to diagnose and correct risky pedestrian behaviors among Chinese children when accompanied with individual debriefing and repetitive practices. School location and grade level had no significant influence on children’s pedestrian performance and learning outcomes, indicating the ubiquity of the pedestrian safety problem and the need for more effective instructional interventions.研究目标：中国儿童行人致死率居高不下，这凸显了在小学推行更具实效的行人安全培训项目的必要性。本研究旨在探究以虚拟现实（Virtual Reality）作为教学技术，识别并改善中国儿童危险行人行为的应用效果。研究方法：本研究招募湖北省三所小学的79名1至3年级学生作为受试者，依据学校所在区域将其划分为城区组（n=20）、随迁子女组（n=29）与农村组（n=30）。所有受试者参与包含三项过街挑战任务的VR教学项目，该项目用于评估五项行人行为指标。受试者首先在首次测试（T1）中独立完成挑战，随后接受个性化复盘指导环节，再进行第二次挑战测试（T2）。本研究以学校区域与年级水平作为被试间因素，对比两次测试的行人表现，并采用归纳分析法解析危险行人行为背后的动因。研究结果：本项目中共观测到三类危险行人行为：冲向马路、在闪烁绿灯时过街以及未观察来往车流。此类危险行为的潜在诱因包括缺乏道路交通标志与交通规则知识，以及日常缺少成人监管。从T1到T2，受试者的整体行人表现均有所提升，且效应量处于中等水平（Ƞp²=0.59，P<0.001）。针对三类危险行人行为，两次测试均呈现出显著的主效应（所有P值均<0.001）；但在所有单变量检验中，测试与区域的交互作用、测试与年级的交互作用均不显著（所有P值均≥0.05）。研究结论：搭配个性化复盘指导与重复练习的VR技术，可有效诊断并改善中国儿童的危险行人行为。学校区域与年级水平对儿童的行人表现与学习成果均无显著影响，这表明行人安全问题具有普遍性，同时也亟需推行更具实效的教学干预措施。