Cell phones and other mobile devices have become part of human culture and are changing human activity and lifestyle patterns. Mobile phone technology continuously evolves and incorporates more and more sensors for enabling advanced applications. The latest generations of smartphones incorporate global positioning system (GPS) and wireless local area network (WLAN) location finding modules, vision cameras, microphones, accelerometers, temperature sensors, etc. The availability of these sensors in mass-market communication devices creates exciting new opportunities for data mining applications. In particular, healthcare applications exploiting built-in sensors are very promising. This chapter reviews different aspects of human activity recognition, including review of state-of-the-art technology, implementation, and algorithmic aspects. With the advent of miniaturized sensing technology, which can be bodyworn or integrated in mobile devices, it is now possible to collect, store, and process data on different aspects of human physical activity. This data can enable automated activity profiling systems to generate activity patterns over extended periods of time for, e.g., health monitoring. Collection of activity patterns is dependent on recognition algorithms that can efficiently interpret body-worn sensor data.
Existing activity recognition systems are constrained by practical
limitations such as the number, location, and nature of used sensors. Other issues include ease of deployment, maintenance, costs, and the ability to perform daily activities unimpeded. Sensor outputs might vary for the same activity across different subjects and even for the same individual. Errors can also arise due to variability in sensor signals caused by differences in sensor orientation and placement, and from environmental factors such as temperature sensitivity.
This chapter (1) reviews different reported methods addressing human activity recognition or classification problem, (2) analyzes implementation aspects on smartphones, and (3) suggests advanced algorithms. The content is based on the papers published by the authors in Refs. 1–3.
Online access to SPIE eBooks is limited to subscribing institutions.