Automatic fall monitoring using floor vibration

This work investigates the possibility of monitoring the activity and the falls of people in dwellings using three or more accelerometers fixed on the ground. The main difference between the proposed method and existing ones is the use of acceleration to estimate the impact force by using the apparent mass of the floor; the latter is experimentally identified in each room in which the tests were performed using the heel drop test. The study has two parts: 1. the apparent masses of different dwellings' floors have been measured. 2. the ground reaction force is studied using a purposely designed force platform with a surface of approximately 2 m × 1 m. The force platform allowed the measurement of the forces generated by the falls of 21 subjects, of a crash test dummy (falling in front or rear direction from seated and standing position, with or without the interposition of objects on the trajectory), and of common objects (e.g. dishes, water bottles, books). The impact location is estimated by triangulation, using a wavelet algorithm derived from the existent literature. The results show the possibility of identifying the presence of subjects inside the room and the fall of subjects in the majority of dwellings. We conclude that the proposed method allows a clear distinction between the fall of subjects and objects, given that the difference in terms of force (which is estimated from the floor's apparent mass and from the measured acceleration) is at least of one order of magnitude.