Floor Based Sensor System: Additional Intelligence, Gait Estimation, and Scavenging Charging Characteristics

In this paper we propose further in-depth analysis to our smart carpet, a floor based personnel detecting system. We have added more intelligence by enhancing fall detection algorithms. Both a convex hull, and heuristic algorithms were developed to detect falls. The proposed algorithms detected fall with 95% sensitivity and 85% specificity when combining both methods exclusively. We extracted and estimated gait parameters, comparing our system to the GAITRite system, which is used as gold standard; here we investigate whether the differences between the two systems are statistically significant. The Statistical T-Test showed excellent agreement between the smart carpet and the GAITRite in estimating gait parameters. With (P = 0.55), the walking speed differences of the two systems are not statistically significant. Additionally we studied the characteristics and the behavior of the sensor’s scavenged signal. We designed and built a single large sensor, where subjects performed multiple walks on the sensor, and their data recorded and studied. The sensors’ voltage waveforms behaved differently corresponding to different people and set of walking trials. The covering material, and the environmental conditions affect the behavior of the scavenged signal. More detailed study and experimental trials are needed.