Real-time Motion Classification using Wearable BAN


Human motion classification is emerging technology which can improve healthcare system and to realize context-aware body area network (BAN). This study focused the fact that the radio channel characteristics between sensor devices and coordinator may have stronger potential for motion classification than the conventional methods using accelerometers and video sensors. This page presents a real-time motion classification system using a cheap off-the-shelf sensor network system (TWELITE, Mono Wireless Co.) based on the python machine learning package. The demonstration shows the real-time automatic motion classification for 6 motion scenarios using the time-domain features that are obtained from the radio channel variation between four sensor devices and coordinator.

Related Publications

Y. Ichikawa, M. Kim, “Investigation of Feature Reduction for Body Motion Identification using Radio Channel Characteristics in Wireless BAN,” IEICE Technical Report, MICT2018-4, May 2018 (in Japanese).

M. Kim, Y. Ichikawa, “Simulation-based Body Motion Classifier Using Radio Channel Characteristics,” The 12th International Symposium on Medical ICT (ISMICT 2018), Mar. 2018.