Title:
AUTONOMOUS MOBILE APPARATUS, LEARNING APPARATUS, ABNORMALITY DETECTION METHOD AND PROGRAM
Document Type and Number:
Japanese Patent JP2021110979
Kind Code:
A
Abstract:
To detect or predict an abnormality occurrence state quickly and correctly by an autonomous mobile apparatus.SOLUTION: An autonomous mobile apparatus 1 includes a control unit 1a and a sensor group 1b. The sensor group 1b detects a current state in the autonomous mobile apparatus 1. The control unit 1a acquires time-series data (current sensor data) from a detection start time point detected at the sensor group 1b to a current time point. The control unit 1a accepts input of a plurality of graphs generated based on current division data obtained by dividing the time-series data at a first predetermined interval and the current sensor data into a leaned model, and obtains a result in which an abnormality state in the autonomous mobile apparatus 1 is detected or predicted. The learned model is constructed by inputting the plurality of graphs into a not-leaned model to perform machine learning. The plurality of graphs is generated based on the time-series data detected at an internal sensor group and accumulated division data obtained by dividing the time-series data at a second predetermined interval regarding at least one state of the autonomous mobile apparatus 1 and the same type of another autonomous mobile apparatus.SELECTED DRAWING: Figure 1
More Like This:
Inventors:
KAMATA NORIHIKO
KIDA KOJI
FURUHAMA NAOKI
KIDA KOJI
FURUHAMA NAOKI
Application Number:
JP2020000463A
Publication Date:
August 02, 2021
Filing Date:
January 06, 2020
Export Citation:
Assignee:
NEC COMMUNICATION SYST
UNIV KAGAWA
UNIV KAGAWA
International Classes:
G05D1/02; G06N20/00; G05D1/10
Domestic Patent References:
JP2019123337A | 2019-07-25 | |||
JP2018124639A | 2018-08-09 | |||
JP2011059790A | 2011-03-24 |
Foreign References:
WO2019244930A1 | 2019-12-26 | |||
WO2015141221A1 | 2015-09-24 | |||
US20190310651A1 | 2019-10-10 |
Attorney, Agent or Firm:
Ken Ieiri