2022 年 28 巻 68 号 p. 510-514
This study revealed proper parameters to predict walking turns to-support the residents as a built environment function. We used a non contact sensor that did not disturb everyday life and aimed to extract highly versatile prediction parameters by considering the movement strategy and the rotation direction. After carefully examining the prediction accuracy for the center-of-gravity shift and the head and upper body rotation parameters, we found the position of the head’s center-of-gravity (Left-Right) and the head rotation (Yaw) were valuable indicators. The time needed for prediction was between 0.18 to 0.48 steps with a prediction accuracy of 70%.