Pattern Recognition in Human Motion for Kinetic Energy Harvesting

Abstract

Kinetic energy harvesting from external vibrations is a common technique to scavenge energy from the environment to power a miniature autonomous sensors. In the first generation of energy harvesters, the designers relied on periodic motion to design and optimize the operation of a harvester. Since the functionality of sensors and the types of environment where they can be placed vary significantly, new techniques to scavenge kinetic energy from irregular motion, in particular the one produced by humans, have emerged. In this paper, we study patterns and self-similarity of acceleration time series generated from human motion in application to a technique known as near-limit kinetic energy harvesters. We show that human motion corresponding to specific physical activities such as running or walking possesses high self-similarity and hence is particularly suitable to be used with the near-limit energy harvesting technique.