Anomaly detection of care robot users in standing up using CoG candidates

Anomaly detection of care robot users in standing up using CoG candidates

Blog Article

There is a growing demand for robots that can assist the elderly to stand up independently.To this end, it is important for the robots to estimate the user's condition and provide appropriate assistance.In particular, the Horse Summer Sheet Liners robots are required to be able to detect abnormal conditions and prevent accidents.However, accurate measurement of human posture requires a large number of sensors, making the system complex and difficult to handle.To estimate the state of a care robot user using a small number of sensors, we have proposed a method to calculate the position of the center of gravity as candidates.

However, because the estimation was performed simply by using machine learning, it did not sufficiently consider how the center of gravity candidates changes depending on the state of the robot user.Therefore, it was necessary to measure not only data of normal operation but also data of abnormal states for binary classification.In this study, we propose and validate a method to detect abnormalities in standing without anomaly state data by analyzing how the center of gravity candidates tends to change over time in normal and abnormal standing.The analysis shows that the maximum value of the candidate center of gravity in the forward direction changes sharply when abnormal standing occurs.Therefore, anomaly detection was performed using the value of the second-order derivative of the maximum value of y of the candidate center Dryer Drum Belt of gravity, and abnormal standing was correctly detected.