@techreport{oai:ipsj.ixsq.nii.ac.jp:00228950,
 author = {坂倉, 健太 and 小中, 英嗣 and Kenta, Sakakura and Eiji, Konaka},
 issue = {30},
 month = {Nov},
 note = {2000 年代以降，自律型移動体を利用した警備システムの実用化が始まっており，警備分野においては今後ますます自動化が普及すると考えられる．本論文では警備の中でも巡回警備に着目する．巡回警備で考慮すべき点として訪問間隔の長さと経路の規則性がある．長い訪問間隔はその地点の警備が手薄となることを，経路が規則的であることは予測されやすいことを意味し，両方が揃うと事故や侵入の可能性が高まる．この観点から警備経路計画はある程度不規則であることが望ましいが，どういった不規則性が警備として適切なのかを判断するためには侵入者のモデルが必要となる．本研究では，侵入者を設備や商店に損害を与える人物と定義する．侵入者はある地点での警備の時刻を一定期間観測できること，およびその観測に基づき次の訪問間隔に対する何らかの予測を行うものと仮定する．この仮定の下，カーネル密度推定を用いて侵入者モデルを構成し，それに基づき巡回経路を評価する手法を提案する．, Since the 2000s, security systems using autonomous mobile vehicles have begun to be put into practical use, and automation is expected to become increasingly widespread in the security field. In this paper, we focus on patrol security. The two important factors to be considered in patrols are the length of the intervals between visits and the regularity of the patrol routes. A long interval between visits means that the security at that location will be inadequate, and a regular route means that the patrol timing is predictable. From this perspective, it is desirable for security route planning to be random to some extent, but an intruder model is needed to determine what kind of randomness is reasonable for secure patrol. In this study, an intruder is defined as a person who causes damage to a facility or store. The intruder is assumed to be able to observe the timing of patrol at a location for a certain period of time and to make predictions about the next visit interval based on this observation. Under these assumptions, we construct an intruder model using kernel density estimation and propose a method to evaluate patrol routes based on the model.},
 title = {カーネル密度推定を用いた侵入者のモデルの構成},
 year = {2023}
}