State of the Art - Smartphone-Based Human Activity Recognition - page 34

Chapter 3
State of the Art
3.1 Introduction
The rise of ubiquitous computing systems in our environment is engendering a strong
need for novel approaches to human-computer interaction. HAR systems can effec-
tively contribute to develop these approaches due to the unobtrusive and pervasive
ways of interaction they provide Schmidt et al. ( 1999 ). Human intervention is needed
in many systems for decision making, usually by means of interaction through tradi-
tional devices such as keyboards, remote controls, switches, or touchscreens (Weiser
1993 ). These mechanisms of interaction are becoming intractable considering the
amount of devices we are exposed to every day.
Moreover, we are now facing a new challenge as a result of the easy access to vast
amounts of information coming frommultiple sources. They can contribute to reduce
our demanding interaction with machines to a minimum. This is possible especially
because we are always (involuntarily or not) providing feedback to the environment
through our behavior and actions. Novel systems could therefore become more cog-
nitive and fundamentally transform our ways of interaction with them (Cook and
Das 2012 ). For instance, in the medical field, monitoring physiological signals in
humans, such as an increase in our average daily heart rate or body temperature, can
indicate an emerging health condition. This detection can be possible even without
the patient being aware of the situation or the need of a check up visit to the doctor.
HAR is an active research field in which methods for understanding human behav-
ior are developed by interpreting attributes derived from various sources (Karantonis
et al. 2006 ; Lukowicz et al. 2004 ; Poppe 2007 ) (e.g. by sensing motion, location,
physiological signals, weather and temperature etc.). It aims to identify the actions
carried out by a person given a set of observations of him/herself and the surrounding
environment. HAR has provided substantial contributions in human-centered areas
of study such as AmI, Pervasive Computing (PerComp) and AAL. These areas make
use of HAR systems as an underlying technological tool which gathers behavioral
information from people about their actions and environment during the course of