Analysis of Activities based on a Single-Shoe Equipped with Force Sensors

As networked sensors become ubiquitous and integrated into the Internet-of-Things, new sensor applications provide increased safety to the public. We explore the concept of classifying personal activities using only a single shoe equipped with a low-cost constructed pressure sensor. We envision our wearable to assist in the wide-scale monitoring of citizen activity to improve smart city analytics. The pressure sensor is a polymeric foil encased with aluminum foil that provides stable readings of applied pressure. With only two detection regions: heel and toe, we assess the classification performance using a real-time implementation of a trained machine learning model. We collect training data on two categories: pedestrian movement and sports activities. We achieve an accuracy of 76%.