無人運転列車のためのマルチセンサー物体検出

カテゴリ: 全国大会

論文No: 5-158

グループ名: 【全国大会】令和7年電気学会全国大会論文集

発行日: 2025/03/01

タイトル(英語): Multi-Sensory Obstacle Detection for Driverless Trains

著者名: パンデジョスナ（日立製作所）,清水拓（日立製作所）,小田篤史（日立製作所）,鈴木基也（日立製作所）

著者名(英語): Jyotsna Pandey (Research & Development Group),Taku Shimizu (Research & Development Group),Atsushi Oda (Research & Development Group),Motonari Suzuki (Railway Systems Business Unit)

キーワード: 自動運転|鉄道|障害物検知|センサフュージョン|カメラ|LiDAR|Driverless Operation|Railway|Obstacle Detection|Sensor Fusion|Camera|LiDAR

要約(日本語): This paper introduces a multi-sensory obstacle detection system designed for driverless trains. The system integrates three detection logics-stereo cameras, machine learning (ML), and LiDAR-enhanced by a sensor fusion algorithm to provide accurate obstacle detection under operational conditions. Its performance is evaluated using real-world data collected over a span of 48,000 km, encompassing diverse environments such as curved, gradient, and varying track conditions. The system demonstrates reliable object detection capabilities up to 600 meters, showcasing its effectiveness in ensuring safety and efficiency in autonomous train operations.

本誌掲載ページ: 252-253 p

原稿種別: 日本語

PDFファイルサイズ: 352 Kバイト