The development of robots supporting frequently performed tasks such as standing up and walking is progressing as elderly people with weakened muscles often require assistance in their daily lives. In order to reduce the burden on caregivers, robots must be able to automatically perform supportive actions. To achieve this, it is necessary to estimate the condition of the elderly person using the robot and provide assistance appropriate to that condition. Currently, the research team has proposed a method for calculating the candidate positions of the center of gravity of the robot user and estimating their physical state using machine learning. However, this method requires the acquisition of the user's movements during use of the robot as teacher data, and in particular, the measurement of data for abnormal conditions (such as when the user is about to fall over while walking) could be burdensome for the elderly.

To solve this problem, the research team developed a method to generate candidate center-of-gravity teacher data without measuring movements by using a human body link model. The human body link model is a simple model representing the human body with rigid links and rotating joints. This model can be used to simulate human positions in various states such as sitting and standing, as well as abnormal states, and in this way generate training data. Experiments have confirmed that the care robots can learn based on the training data created by this method, estimate the user’s state, and support standing up, walking, and sitting down actions.