Date of Award
Spring 2017
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Mathematics, Statistics and Computer Science
First Advisor
Ahamed, Sheikh Iqbal
Second Advisor
Smith, Roger O.
Third Advisor
Povinelli, Richard J.
Abstract
Human activity recognition is a technology that offers automatic recognition of what a person is doing with respect to body motion and function. The main goal is to recognize a person's activity using different technologies such as cameras, motion sensors, location sensors, and time. Human activity recognition is important in many areas such as pervasive computing, artificial intelligence, human-computer interaction, health care, health outcomes, rehabilitation engineering, occupational science, and social sciences. There are numerous ubiquitous and pervasive computing systems where users' activities play an important role. The human activity carries a lot of information about the context and helps systems to achieve context-awareness. In the rehabilitation area, it helps with functional diagnosis and assessing health outcomes. Human activity recognition is an important indicator of participation, quality of life and lifestyle. There are two classes of human activities based on body motion and function. The first class, simple human activity, involves human body motion and posture, such as walking, running, and sitting. The second class, complex human activity, includes function along with simple human activity, such as cooking, reading, and watching TV. Human activity recognition is an interdisciplinary research area that has been active for more than a decade. Substantial research has been conducted to recognize human activities, but, there are many major issues still need to be addressed. Addressing these issues would provide a significant improvement in different aspects of the applications of the human activity recognition in different areas. There has been considerable research conducted on simple human activity recognition, whereas, a little research has been carried out on complex human activity recognition. However, there are many key aspects (recognition accuracy, computational cost, energy consumption, mobility) that need to be addressed in both areas to improve their viability. This dissertation aims to address the key aspects in both areas of human activity recognition and eventually focuses on recognition of complex activity. It also addresses indoor and outdoor localization, an important parameter along with time in complex activity recognition. This work studies accelerometer sensor data to recognize simple human activity and time, location and simple activity to recognize complex activity.