Abstract: 　To solve the problems related to pedestrian traffic behavior simulation and behavioral data acquisition in public buildings, a relevant platform was established based on existing pedestrian traffic theory. To do this, virtual simulation and human-computer interaction technologies were used. A preliminary experiment involving pedestrian traffic data acquisition was carried out using this platform. Firstly, an experimental platform for virtual pedestrian simulation and data acquisition was constructed based on the Unreal virtual engine using an HTC vive helmet, Omni universal treadmill, and other equipment. Secondly, a total of 23 groups of path-finding experiments were carried out in the virtual environment of Yantang metro station in Guangzhou. Behavioral data from respondents, such as the timestamps, space coordinates, and sightline angles (straight/up/down) were recorded. Finally, the collected data were used to analyze the route data and behavioral data of pedestrians, the distribution of routes taken by different respondents to reach the same destination, the length that the respondents walked together, the speed and variance of bottlenecks, etc. The results indicated that:
　　(1) The proposed platform based on a VR system is feasible. The questionnaire showed that more than 95% of respondents confirmed that the system had a high degree of authenticity and was a good imitation of the real scene. The relevant experimental results are in line with the behavioral characteristics of pedestrians in the actual environment.
　　(2) On this platform, different scenes and pedestrian flows can be set for different types of pedestrian traffic behavior simulations to obtain detailed pedestrian data at any time interval, allowing the researcher to conduct different types of analyses.
　　(3) In this paper, a virtual test environment was studied which enabled natural walking and panoramic viewing. A new platform and method for the study of pedestrian traffic behavior in public buildings is provided by recording data on pedestrian activities and visual trajectories for analysis and quantitative research.
　　(4) The behavioural patterns of respondents in terms of path-finding can be reflected by recording their sightline changes (straight/up/down). The sightline remains mainly in the horizontal direction and serves as the basis for further study of environmental behaviors.

Key words: virtual reality, public buildings, pedestrian traffic behavior, data collection and evaluation

摘要： 为解决公共建筑中个体行人交通行为的模拟，行为数据采集等相关问题，在已有行人交通理论研究成果的基础上，应用虚拟仿真技术及人机交互技术，建立行人交通行为模拟、行为数据采集平台，并在此平台基础上开展了初步的行人交通数据采集实验。首先采用Unreal虚拟引擎，HTC Vive头盔、Omni万向跑步机等设备搭建了虚拟行人模拟与数据采集实验平台；然后以广州地铁燕塘站为例，进行了23组虚拟环境寻路实验，记录了测试者在不同时刻位置，视角变化等信息；最后利用采集数据对行人交通行为进行了相关分析。结果表明：基于VR系统的行人仿真及交通行为数据采集平台具有可行性，相关的实验结果符合行人在实际环境中的行为特征。通过对行人视角变化记录数据的分析，可以对行人细节交通行为进行判断，并作为进一步环境行为研究的基础。

关键词: 虚拟现实, 公共建筑, 行人交通行为, 数据采集与评价