A baseline walking dataset exploiting accelerometer and gyroscope for fall prediction and prevention systems

Fall datasets usually record normal activities and transitions from one posture to another one with falls. Many fall detection datasets based on different sensors are adopted by researchers to improve their systems. Although fall avoidance are dramatically increasing, a public fall prediction and prevention dataset based on an accelerometer and gyroscope is absent. So, this study creates a dataset based on the state-of-the-art techniques in simulating a fall. Different techniques are evaluated to find the best fall simulation. Since accelerometer and gyroscope sensors embedded in a smartphone are recognized to be suited for fall avoidance systems, in this study, they are used to obtain data from users. At the end, some statistical analysis of the observed data are presented and a nonlinear regression model is proposed.