Activity Recognition With Smart Polymer Floor Sensor: Application to Human Footstep Recognition

Smart systems are commonly used in our environment and satisfy numerous applications for the end user. However, the need for unobtrusive systems is constantly growing as we are living in a world where such systems are more common place and pervasive than ever. More specifically in the field of smart flooring systems, being subtle and unobtrusive is key as people come into intimate contact with these systems. The smart flooring system that has been developed is reliable, persistent, and completely transparent to the user, therefore making it ideal for many applications including safety, intelligent surveillance, patient monitoring, injury prevention, asset tracking, and entertainment. The work presented herein reports a novel approach to identify a footstep pattern from a single sensor area. Signals are processed according to a normalization calculation process and correlated with signal references through the use of a single piezoelectric sensor. Several correlation methodologies were performed by comparing an incoming signal with a signal reference to generate coefficients. These coefficients can be thought of as similarity indicators and have allowed us to distinguish between events, such as human footsteps among other events with 100% accuracy.