Part SLAM: A Succinct and Discriminative Method for Scene Matching via Unsupervised Part-based Modeling

2017 Fiscal Year Final Research Report

• Project/Area Number: 26330297
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Intelligent robotics
• Research Institution: University of Fukui
• Principal Investigator: Tanaka Kanji 福井大学, 学術研究院工学系部門, 准教授 (30325899)
• Project Period (FY): 2014-04-01 – 2018-03-31
• Keywords: 視覚移動ロボット / 自己位置推定 / 地図作成 / 情景部品モデル / 深層学習 / 変化検出

Outline of Final Research Achievements

In this study, we realized a next generation SLAM (online map learning) technique named ``part SLAM". More formally, we are based on a light-weight and high-accuracy compressive map representation ``unsupervised scene part model" and developed the new SLAM technique. Furthermore, we developed a versatile SLAM system based on monocular camera and verified efficacy of the developed system in a challenging problem called ``cross-season visual place recognition".

Free Research Field

視覚移動ロボット