|
|
The
research challenges involved in development of smart attire are
multi-faceted. As Computer Science researchers, we are highly excited
about the research challenges that we will face during the course of
this project.
The work on creating a prototype of smart clothing began as a
motivation to augment the smart in-home monitoring system developed by
the Medical Automation Research Center (MARC) at the University of Virginia.
The smart in-home monitoring system targets to improve the quality of
health care and the quality of life of elder people, by providing a
smart in-home monitoring system composed of a suite of low-cost
non-invasive sensors, and a data logging and communications module, in
addition to an integrated data management system, linked to the
Internet. Smart Attire was conceived to fill the gap created by the
absence of sensing, data logging, and communication capabilities
outside the home environment.
We
have been working on the development of a research prototype of a piece
of smart clothing for the past year and a half. Our work has resulted
in a Smart Jacket
that is
built by weaving MicaZ motes in the lining and padding of a heavy
winter jacket. MicaZ motes are sensor nodes that are half the size of
modern cell phones and are equipped with processing, storage, sensing,
and communication capabilities.
The initial prototype uses a heavy winter jacket, as the current form
factor of MicaZ motes makes it feasible to embed them only in a heavier
piece of clothing. But, the decreasing trend in the size of these
devices suggests that, we will soon be able to embed these devices in
lighter clothing like shirts, pants, and windsheeters. Our jacket
prototype is capable of monitoring the motion and
location information of a person remotely using accelerometers (devices
that measure acceleration) and GPS sensor, respectively. A typical
operational scenario involves a person wearing the jacket and
performing outdoor activities. The jacket records the motion and
location
information in the flash memory of the MicaZ motes. Upon coming in the
range of a base station (a mote attached to the PC), the data collected
is uploaded to the PC, transparently.
The idea of clothing with sensors and computing devices embedded in
them is very exciting. In striving to do so, one of the fundamental
problems is to conceive an architecture that will ease the
development of software for such clothing. The motivation for
development of such an architecture is three-fold. To begin with,
various sensors can be introduced by the manufacturer of clothing,
hence a platform for easy incorporation of these sensors is required.
Second, several garment instances must be able to leverage the same
architecture. For example, different clothing items (like shirts and
pants) should be able to work together in harmony. Finally, it is
important to have multiple data interpretation algorithms to identify
various activities performed as well as multiple user interfaces, that
provide different visualizations of the same data based on the choice
of the person. To accommodate these needs, we designed a new
general
system architecture that allows flexible and modular development of software for smart attire. This generalized system architecture is shown in the Figure below.
Figure 1: General System Architecture
| Future Research Challenges |
Smart attire poses research challenges that are multi-faceted and
encompass various fields of Computer Science and Electrical
Engineering. Some of these challenges are enumerated below, with a
brief explanation provided for each of them:
- Hardware: Development
of hardware that will be specifically oriented to smart attire is an
important problem that needs to be addressed. Egs: Sensor nodes that
can sense vital signs of humans, computing devices that are easily
washable, devices that scavenge power from human motion.
- Ipod Integration:
One of the recent ideas we had was to integrate smart attire with
ipods. The widespread use of ipods makes it an attractive option for us
to integrate our initial prototype with these music players. We are
exploring the realm of possibilities and improvements that we can
achieve with a better processing and storage capabilities.
- Operating System and Middleware Services: Several
operating systems exist, that vary from standard PC OSes like Windows
XP and Linux to embedded OSes like TinyOS and LynxOS. But, a crucial
component that is missing in adapting these operating systems for smart
attire is that, they are not context aware. These operating systems run
on machines with which humans interact. Contrary to this, the operating
system for smart attire needs to be aware of the activities of the
person wearing such clothing. Middleware services that Identifying the core services that form
the building blocks of an operating system for smart attire is a very
interesting and challenging problem.
- Human Activity Identification:
Several researchers have delved into the problem of identifying human
activities, most of these use cameras to achieve their goal. Work has
been done in identifying activities using accelerometers. But, the
problem of identifying complex human activities (such as eating,
drinking a cup of coffee) using accelerometers is still an open
research area. Development of a generalized framework for identifying
human activities is a very challenging and thought provoking question.
- Privacy and Security: The
prime concerns with any system that collects sensitive information
regarding humans are privacy and security issues. Novel privacy and
security solutions are required to address these issues.
- Data Mining: A
wealth of information is collected by recording every moment of the
life of a person. Organizing and querying such data is indeed a very
interesting issue. Extracting patterns out of these data is crucial for
the system's functionality. For example, studying the long term
ambulatory motion of a person may help identify the onset of certain
diseases in their preliminary stages.
|
