Toward social approach of classifying road lighting situation for community-centric system

This paper proposes a method for classifying roads' lighting situation at night based on social approach. Information about roads' lighting situation is important in terms of safety and security of local communities, which is used for discussing anti-crime activities and route recommendation at night. However, it is difficult to know actual lighting situation on roads because of the lack of information about streetlamps and the existence of other light sources and obstacles. In order to tackle this problem, this paper employs social approach, in which local residents collaboratively collect roads' lighting situation with their smartphones. As fundamental technology for the social approach, a classifier that judges road's lighting situation into 3 levels is build based on three attributes, which are calculated from illuminance data collected with a smartphone. An experimental result using 164 actual road data shows the accuracy of the classifiers is over 0.860 in the best case. This paper also discusses challenges for realizing the social approach such as the existence of various smartphones with different characteristics of illuminance sensors.