2016 年 82 巻 840 号 p. 15-00689
This paper introduces a distance-based observation model for standing and sitting behaviors by using a proximity sensor as an interface function toward the development of a novel lower-limb assist robot. The proposed observation model employing relative distances to each part of a human body enables to determine a behavioral state transition from sitting to standing, or conversely. When the observation model capable of monitoring these behaviors continuously is applied to the assist robot, it would be very convenient to employ the device due to not requiring any manual control devices. Specifically, the observation model without visual data will be a great help to decrease mental burdens such as privacy. Practically, to verify the effectiveness of the observation model, the evaluation experiments were conducted by 30 subjects including 17 persons over the age of 60 and the results are described in detail. As a result, we could confirm that the observation model provides the assist robot with reliable recognition without the use of any manual controls or wearing a specific device.