2017 年 16 巻 2 号 p. 105-112
The rapid spread of smart phone has caused the texting-while-walking problem. Earlier studies use acceleration sensors to control the phone operation while walking and issue warning signs, but sometimes wrongly determined because the unconscious behavior of users and the influence by different environment. In this study, in order to improve the accuracy of walking determination, we analyzed acceleration data under different smart phone use conditions, and developed an application to issue a warning when the person is using a phone while walking under various environments. As result of evaluation, developed application was able to detect the use of the phone while walking at a high accuracy level (95-98%), and reduce false detection when not walking (0-5%).