Detecting lost tourists using a learning method based on behavior history and geospatial information

2023 Fiscal Year Final Research Report

• Project/Area Number: 21K12140
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 62020:Web informatics and service informatics-related
• Research Institution: Osaka Seikei University (2022-2023); Kyoto University (2021)
• Principal Investigator: Kasahara Hidekazu 大阪成蹊大学, データサイエンス学部, 教授 (00784191)
• Project Period (FY): 2021-04-01 – 2024-03-31
• Keywords: 観光情報学 / 知能情報学 / 移動軌跡分析

Outline of Final Research Achievements

The research plan comprised three main objectives: 1) improve deviant behavior detection using a mixed Gaussian model, 2) develop offline detection methods utilizing behavioral history and geospatial information, and 3) expand these methods to an online platform. The initial phase encountered challenges due to the pandemic, preventing the collection of the student dataset. Consequently, the focus shifted to studying low mountains where beginner hikers frequently get lost, and a hiking trajectory dataset was obtained from a company.
During FY2022-3, a new method was introduced to link the difficulty level of mountain areas with hiker proficiency. This involved analyzing the continuation of climbing activities across a mesh of varying difficulty levels; if a hiker's activity persisted in an area exceeding their skill level, it was considered a lost trail. This approach yielded promising results, providing a valuable framework for enhancing safety and navigation in mountainous areas.

Free Research Field

知能情報学

Academic Significance and Societal Importance of the Research Achievements

本研究は、観光客が道迷いに気づく前に、システムが道迷いを検出する手法の開発を目指した。人は天気や混雑といった実時間情報に応じてダイナミックに行動や目的地を変化させるので、目的地への最短経路からの逸脱を異常と見做す手法では道迷い検出にはそぐわない。また、意図した「寄り道」と意図しない「道迷い」を区別することが難しいという問題もある。本研究は人の意図しない行動を検出する試みであり、パターン認識における行動意図推定の発展系と位置づけられる。学習型異常検知手法に地理空間情報を織り込む手法の開発にも繋がっており、学術的に重要な研究である。自動運転や介護といった分野への波及効果も期待できる。