This study develops a new method to estimate real-time pedestrian comfort, i.e., Pedestrian Level of Service (PLOS). Interest in pedestrian comfort in urban spaces has been increasing. Many studies have focused on investigating the key factors that define and affect PLOS. Definitions of PLOS vary. Some studies used the concept of vehicular LOS to define PLOS by using pedestrian volumes and average pedestrian walking speed, etc. Others used a more qualitative definition based on survey data collected from targeted pedestrians. A few studies have considered individual pedestrian gait characteristics. This study proposes a novel computer vision-based technology to measure PLOS on a real-time basis. The system uses advanced pedestrian tracking technology to automatically detect and analyze individual pedestrian gait characteristics from CCTV camera videos. Our system, called “3DTown,” tracks and renders pedestrians with 3D avatars in highly-detailed static and dynamic environments. A major walkway corridor on York University’s Keele Campus in Toronto was used as a testbed for the system. Two Pan Tilt Zoom (PTZ) surveillance cameras were used for measuring pedestrian gait characteristics such as step length, walking speed, walking acceleration or deceleration, deviation from walking path, and upper body motion. A fuzzy inference system was then applied to analyze these characteristics and produce a novel measure of pedestrian comfort for our study area. The proposed method is also applicable to more complex walking environments such as subway platforms and sport stadiums. As the system can automatically and quantitatively analyze the level of pedestrian comfort on a real-time basis, it should be possible to use the system’s outputs to initiate further improvements in the walking environment.
