To make it possible to classify whether information representing evading navigation is contained in time-sequential data representing movement of a moving entity.SOLUTION: An evading navigation decision system 100 includes classification learning means 16 that learns an evading navigation classification model, which is used to classify time-sequential data in terms of evading navigation, using classification teacher data, which is a combination of a feature vector, which is extracted as information concerning a collision of a moving entity that is an object, and evading navigation information, which signifies whether the moving entity has undergone evading navigation, on the basis of the time-sequential data relating to navigations of plural moving entities, classification prediction means 18 that uses the learned evading navigation classification model to classify the time-sequential data of the moving entity in terms of evading navigation, and evading navigation classification prediction result output means 20 that outputs a result of classification of the moving entity in terms of evading navigation.SELECTED DRAWING: Figure 1
【課題】移動体の移動を示す時系列データに避航を示す情報を含むか否かを分類することを可能にする。【解決手段】複数の移動体の航行に関連した時系列データに基づいて、対象とする移動体の衝突に関する情報として抽出された特徴ベクトルと、移動体が避航したか否かを示す避航情報とを組み合わせた分類用教師データを用いて、時系列データを避航分類するための避航分類モデルを学習する分類学習手段16と、学習済みの避航分類モデルを用いて時系列データから移動体の避航分類を行う分類予測手段18と、移動体の避航分類の結果を出力する避航分類予測結果出力手段20とを備える避航判断装置100とする。【選択図】図1
EVADING NAVIGATION DECISION METHOD OF MOVING ENTITY, EVADING NAVIGATION DECISION SYSTEM, AND EVADING NAVIGATION DECISION PROGRAM
移動体の避航判断方法、避航判断装置及び避航判断プログラム
2019-10-10
Patent
Elektronische Ressource
Japanisch
Europäisches Patentamt | 2020
|Europäisches Patentamt | 2022
|