Detecting smoothness of pedestrian flows by participatory sensing with mobile phones

Abstract

In this paper, we propose a novel system for estimating crowd density and smoothness of pedestrian flows in public space by participatory sensing with mobile phones. By analyzing walking motion of the pedestrians and ambient sound in the environment that can be monitored by accelerometers and microphones in off-the-shelf smartphones, our system classifies the current situation at each area into four categories that well represent the crowd behavior. Through field experiments using Android smartphones, we show that our system can recognize the current situation with accuracy of 60-78%.

A Survey of Techniques for Automatically Sensing the Behavior of a Crowd

“…[79] Ensures privacy: anonymity, communication security; high concern; [25,58] Incentives No incentives; no discussion; e.g. [64,78] Provides incentives; no details about them or their effectiveness; e.g. [88] Provides incentives; not very e ective; not a primary concern; e.g.…”

“…[58]: the participants start the media recordings Unclear, but may be transparent, e.g. [64]: seems transparent but not discussed, it's based on sound recording so it might require permissions to trigger it e participants are not required to do anything or they are not aware of the collection campaign; e.g. [55] Resource consumption High energy consumption, many resources involved; no sensor sampling strategies; e.g.…”

“…[55] Resource consumption High energy consumption, many resources involved; no sensor sampling strategies; e.g. [64] Unclear, but most likely high, e.g. [55]: collected ba ery levels, observed that the ba ery dropped 20% in 2 h, controlled the submit rate based on the critical places Unclear, but most likely low, e.g.…”

“…e spot-on systems with notable good accuracy, in-depth experiments (varied scenarios, multiple type of mobile devices) and complex analysis (e.g. considering energy consumption, scalability) are Kannan et al [36] for crowd counting, Weppner and Lukowicz [87] for density estimation, Nishimura et al [64] for congestion classication and Kjaergaard et al [41] for ock detection. Regarding the la er, the ocks and other pa erns such as leadership and following, make the topic of the systems described by Kjaergaard et al in [37,38,41,42].…”

Gamification Mechanism for Enhancing a Participatory Urban Sensing: Survey and Practical Results

“…"Road" is one typical application that has high affinity for participatory sensing. Human movements on road [33], [34], road surfaces [35], [36], traffic jams [37], and parking congestion [38], [39] are also popular targets. Although external sensors are required for the participants, air quality [40], [41], [42] and fuel efficiency [43] and scenic roads [7] are also sensed by participatory sensing.…”