The automatic evacuation guiding using mobile devices is a promising disaster-response approach for large-scale disasters. When a disaster occurs, each mobile device recommends a route to a refuge and automatically estimates the state of road segments by comparing the recommended route and observed trajectory. The estimated information is shared among devices using available communication technologies. This scheme assumes that each evacuee follows the recommended route but it might not be satisfied when an evacuee behaves as an outlier taking unexpected movement, which will disturb the normal evacuation. This paper analyzes the effects of outliers and reveals the following characteristics through simulations: The number of passable road segments misestimated as impassable increases with outliers; Many of them, however, can be corrected by passable estimation from other outliers; The remaining misestimation increases the evacuation completion time of 26.5% of normal evacuees by more than 100~s. Since the speedy and accurate outlier detection is difficult in the severe communication environments, this paper alternatively proposes a mechanism that delays to judge a road segment as impassable until gathering the corresponding impassable estimation from M devices. The simulation results show that M=2 achieves a balance between the speediness and safety in evacuation guiding.
Jun Kawahara, Takanori Hara, Masahiro Sasabe, On Robustness Against Evacuees' Unexpected Movement in Automatic Evacuation Guiding, Computers and Electrical Engineering, 105, pp.108531: 1-17, January 2023.
@article{kawahara23RobustnessEvacueesUnexpected,
author = "Kawahara, Jun and Hara, Takanori and Sasabe, Masahiro",
title = "On {{Robustness Against Evacuees}}' {{Unexpected Movement}} in {{Automatic Evacuation Guiding}}",
year = "2023",
month = "January",
journal = "Computers and Electrical Engineering",
volume = "105",
pages = "108531: 1-17",
doi = "10.1016/j.compeleceng.2022.108531"
}