Final
Report of ITS Center Project: Regional pedestrian activity measurement
A Research Project Report
For the ITS Implementation Research
A US DOT University Transportation
Center
REGIONAL PEDESTRIAN ACTIVITY MEASUREMENT (PROJECT 434911)
Principal Investigator:
Aaron Schroeder, Ph.D.
Virginia Tech Transportation
Institute
3500 Transportation Research Plaza
(0536)
Blacksburg, VA 24061
Phone: 540-231-1505
Fax: 540-231-1555
September 2006
Disclaimer
The
contents of this report reflect the views of the authors, who are responsible
for the facts and the accuracy of the information presented herein. This
document is disseminated under the sponsorship of the Department of
Transportation, University Transportation Centers Program, in the interest of
information exchange. The U.S. Government assumes no liability for the contents
or use thereof.
Table of Contents
|
Words of appreciation |
3 |
|
Project overview |
4 |
|
Project objectives |
4 |
|
Project tasks |
4 |
|
Project accomplishments and findings |
5 |
|
Conclusions and recommendations |
18 |
Appendices
|
Appendix A – Communications |
24 |
|
Appendix B – Technologies (product web sites) |
25 |
|
Appendix C – Literature review paper |
26 |
List of Tables and Figures
|
Figure 1 – The Nextel Motorola i605 |
11 |
|
Figure 2 – The Nextel RIM 7100i |
12 |
|
Figure 3 – The HP iPAQ iPAQ hx2495 |
13 |
|
Figure 4 – GT1M Actigraph |
14 |
|
Figure 5 – The Globalsat BT338 |
15 |
|
Figure 7 – The Holux GPSlim236 |
15 |
|
Figure 8 – The Trackstick |
16 |
|
Table 1 – A comparison of initial wearable concepts as
compared to evaluated concepts |
19 |
Words of Appreciation
This effort greatly
appreciates the resources and time the following individuals and organizations
provided.
The U.S. Department of
Transportation, University Transportation Centers Program, and the Virginia
Tech Transportation Institute – for their fiscal and resource support.
Nextel, specifically Helen
Franks and Marselle Culpepper for their willingness to loan us two phones,
associated service, and applications.
Kathy Hosig, Ph.D. and Lisa
Schweitzer, Ph.D. for their invaluable expertise.
Project overview
American society has become
automobile-centric to such an extent that any inquiry into alternative modes of
transportation – those modes that are not dominant – frequently finds little
response as little data or information exists for such modes. The ever-increasing general awareness of the
systemic nature of resources that began in the 1990s spawned an extensive set
of inquiries that attempted to reach into the environmental and health benefits
of alternative modes of transport.
Lacking data on alternative modes, practical inquirers set out to
determine if existing automobile data collection mechanisms could be used for alternative
mode data collection. For example,
transportation professionals sought additional utilization of existing traffic
cameras, capable of counting and classifying automobiles, to also count
pedestrians. Though this approach seems
practical, the assumption that pedestrians follow the same rules and roads as
the automobile is questionable.
To better understand
pedestrian behavior, oral and written surveys
and travel diaries have often been used.
But these can lack accurate spatial characteristics of movement, as the
participant may not convey all information.
Technical data collection has improved this. Wearable technologies that collect the pedestrian’s spatial
location and prompt them to provide a response as to why they have just changed
direction, etc. have been used. While
collecting a tremendous amount of information on a pedestrian’s location, and
their decision-making, they can impede the pedestrian’s progress, as they will
interrupt the pedestrian’s routine with inquiry. However, a large-scale deployment of such wearable tracking
technologies – to collect the whereabouts of a population of participants –
combined with land use and other data sets to provide a comparative commentary
on participants’ movement – may reduce the challenges of a survey of
individuals while providing a significant amount of information to inquirers as
to human movement in a region. Further,
given the nature of funding for organizations interested, and given the
technological capabilities of today’s wearable devices, it may be of great
interest to team between several disciplines, provide one wearable technology
architecture, and glean data relevant to multiple disciplines’ inquiries.
It was the impetus of this
effort to provide a review of those who would be interested in such a study,
and take a quick look at wearable technologies and their associated wearable
architecture arrangements in order to establish the basis for deploying
something of this nature. This effort
revealed that those disciplines interested in human movement through a region are
quite diverse, ranging from public safety to biomotion. However, teaming for an immediate
co-beneficial technology deployment would be those involved in the health and
well-being of a person (for example, physical activity) and those involved in
transportation (for example, transportation planning and engineering).
In terms of wearable
technologies this effort developed and evaluated six wearable architectures and
found that at least two, with some modification, would be worth further
exploration. The critical features in
their use between the researchers and the participant (the wearer) are cost,
the number of devices to carry, and the number of steps involved in device
maintenance. Minimizing all is critical
to the success of a future deployment.
This project’s initial
objectives and tasks follow.
Project objectives
The goal of this project was
to develop a detailed understanding of what would be required to undertake a
wearable-technology pedestrian survey and to establish, as a foundation, the
next step of technology prototyping.
Over the long term, the goal would be to develop and deploy a
wearable-technology pedestrian survey using wireless technologies, Global
Positioning System (GPS) receivers, and clothing-integrated
accelerometers. Such a deployment would
benefit researchers in multiple domains including: human health and medicine,
civic planning, and transportation.
Project tasks
The
tasks as initially drawn out for this effort include those listed below. Many of the tasks were conducted
simultaneously.
·
Review literature on
pedestrian movement: Macro (region-wide) - automated tracking, surveys, or
systems similar to the one described above; Micro (human physiology) relative
to transportation systems and facility design - how may accelerometer systems
help to improve design under 'real-world' survey?
·
Review technologies:
Macro (reviewing portable GPS technologies); Micro (reviewing human-movement
assessment technologies such as wearable accelerometers).
·
Assess the feasibility
of integrating technologies while considering the pedestrian survey
participant: technology integration (wireless connectivity vs. data storage);
human-technology integration; and cost.
·
Identify key
stakeholders who are interested in this area of research and begin to establish
collaborative relationships.
·
Conduct real or virtual
'round-table' sessions with domain experts to determine overall feasibility and
validity - "if such a survey were possible, in what might you be
interested?"
·
Develop prototype technology
concept and survey deployment schemes.
·
Develop final report
synthesizing above findings.
Project accomplishments and findings
Review of the literature
The initial phase of this
effort began with an examination of the international and national interest in
the study of human motion. This
snapshot provided insight into cross-discipline interest and nomenclature. The review culminated in late October of
2005 with a presentation to Association of Collegiate Schools of Planning
(ACSP) conference in Kansas City. The
presentation of the paper, “Pedestrian Activity Measurement: A Review of the
State of the Art and the State of the Practice,” was given in a new session at
the conference, Built Environment and Physical Activity. The new track serves as a nexus of health
and urban planning academics and practitioners. It quickly became standing room only – clearly demonstrating that
the communities of health and urban planning are ready to commit to formally
discussing, and addressing, the issues with land use and physical
wellness. With urban planning’s close
association to transportation engineering, it seems likely that a health,
planning, and engineering dialogue will commence on the same topic.
The initial findings
presented were the highlights of a review encompassing the many disciplines
that have an interest in pedestrian activity measurement. Dominant were the urban planners and
transportation engineers. These disciplines
collect data on pedestrian movement through a region or at the streetscape (for
example, a count of pedestrians crossing at an intersection). Disciplines interested in the physical
wellness of the pedestrian also had representatives who were also very
interested in the effort. Such wellness
of the pedestrian includes: the physical security of the individual, for
example, walking at night with or without street lighting in an area known to
have a high crime rate; the safety of the individual walker, for example, the
ability of the pedestrian to evacuate a building or a region; and the health of
the walker proper (e.g., the motion of the body through space across varying
terrain [biomotion studies]).
The full details of this
research may be reviewed in Appendix C, where “Pedestrian Activity Measurement:
A Review of the State of the Art and the State of the Practice” is presented.
Identification and communication with people and disciplines of interest
Review of the literature
still left a series of practical questions unanswered: If a study were to be
conducted locally and regionally, who might be interested in the movement of
people? Who might be experts in
conducting practical experiments involving people in general? Who might be in need of data or information
on the movement of people in general?
In answering these questions for this effort, an attempt was made to
communicate with local, regional, and statewide actors that might have interest
in seeing a project evolve to capture data on pedestrian and regional human
movement in general. This search began
in the organizations encompassing the Virginia Tech Transportation Institute
(VTTI). This included a search of
Virginia Tech, Blacksburg (the town in which Virginia Tech resides), and the
state of Virginia. It is unusual for a
region to have access to the technological sophistication that Virginia Tech
offered. In this regard, our effort had
a significant advantage over regions that do not maintain such a
University. However, the other institutions,
the inter- and intra-regional actors involved in health, planning,
transportation, safety, pedestrian and bicycle activism, etc. are not
necessarily unique to this region. As
such, any follow-on research or deployment would be able to identify similar
actors in their regions that may wish to partner, or assist, with the research
or deployment of such tracking technologies.
Organizations that may prove
of universal interest to such an activity include:
·
Regional pedestrian,
hiking groups, or bicycling
·
Planning offices
·
Safety offices
·
Engineering offices
·
Transportation offices
·
Health and nutrition
organizations
A complete list of
communications may be reviewed in Appendix A.
Development of initial concepts
Prior research experience in
wearable technologies, a review of associated literature, and communications
with international, national, and local experts led to the development of
several conceptual systems for capturing and analyzing data on pedestrian
location. These systems included a
field and desktop technological component.
The desktop component was similar across all conceptual varieties. The desktop technology included a computer
connected to the internet capable of connecting to the field devices either
directly or through the internet: this desktop system was the central
repository for collected data and could serve also as a data analysis and
sharing mechanism.
The field technology
component for any of the concepts was to be a collection of wearable technologies that afforded information about the
wearer’s location and movement. These
devices were to provide this information with minimal impact on the
wearer. That is, one would affix them
to their person, activate the technologies, and go about their daily
routine. At no point should the wearer
need to reactivate or respond, or otherwise interact with these devices till
the end of their daily routine when they would likely need to perform some kind
of daily maintenance ritual with these devices, such as downloading data or
charging the batteries. The hope being that while worn, the individual wearer
would forget that he or she was participating in a study, thereby reducing user
bias.
In should be noted that there
were two papers that proved critical in the development of these concepts. The papers provided documented examples of
real-world data collection and rapid analysis techniques. Any further investigation into deployment of
such technologies should involve a review of papers and their associated
research.
Oliveira, et al., used two
data-logging field technologies, an Actigraph accelerometer and a Geostats
Wearable Geologger GPS receiver, to capture the location and movement of an
individual.[1] Following data capture, a proprietary data
analysis system, GeoStats Trip Identification and Analysis System (TIAS) was
used to determine the individual’s activity and assign a probability of
transportation mode of the wearer’s daily activity. The technological application of this research proved uniquely
compelling to the development of the wearable concepts for this effort. In their research, an individual wearing
such technologies was capturing travel origin and destination information, and
the technologies that were assigned to collect movement information, number of
steps taken or level of activity, was also being used to associate with the
location information to provide modal information. In one instance, multiple disciplines’ inquiries into the
condition of a pedestrian were capable of being addressed.
Doherty et al. used a
Bluetooth-enabled cell phone in combination with a Bluetooth-enabled GPS
receiver to collect information on an individual’s location in near-real-time.[2] A GPS signal collected by the GPS device was
sent to the cell phone via Bluetooth, and the cell phone, with appropriate
service, then sent that information on and
through the internet to the researcher’s central repository for analysis. Additionally, their design used the unique
capabilities of the cell phone to allow the wearer to respond to a
location-based prompted recall survey.
It was their intention to find out more about travel habits through such
prompted recall. While their
technological sophistication was one that this effort sought to emulate, the
two efforts differed philosophically on the prompted recall. In theirs, they actively captured additional
information from the wearer when that individual approached a specific
location. In our effort’s concept, it
was believed that such prompted recall might interfere with an individual’s
decision-making processes on choice of path or mode, thus, the movement alone, in
combination of other data sources, would provide data for later analysis.
Further communications with
Doherty revealed additional capabilities that had been developed.[3] His team had established a sever which would
accept Geospatial Information System (GIS) files of a region’s roadway network
and land use as well as any GPS tracks collected by an individual wearing
technology as described above. This
information would then automatically generate a probability of the individual’s
transport mode. Thus, the simple
technology that provides a location for the individual, when compared with
additional information, can provide insight into that individual’s movement to
address multiple disciplines’ inquiries.
The analytical capabilities
the aforementioned research efforts were well documented and sound. They demonstrated that there were stable and
accessible applications for data analysis that could accept inputs from a wide
variety of standardized technologies to provide insight into pedestrian
movement. This suggested that our
initial objectives could be more focused; our effort could focus on how to
capture and store the data rather than how to analyze the data. Thus, our efforts focused in on determining
how friendly wearable devices were in terms of cost, comfort, configuration,
and data collection.
The VTTI research effort’s
initial conceptual architectures included:
·
Wearable Concept 1 –
disparate data loggers. In this
architecture there is a wearable accelerometer that is capable of storing some quantity
of activity data (at least a day’s worth at 1-second intervals). There is also a wearable GPS device capable
of storing some quantity of activity data.
In such a case, some maintenance of the devices is required. Data would need to be downloaded and the
device’s power supply replenished. This
concept is similar to that defined by Oliveira, et al.[4]
·
Wearable Concept 2 – a
data storage device wired to two collectors.
In this instance, an accelerometer is attached to a wearable storage
device that is simultaneously connected to a GPS unit. Thus, the participant carries three devices,
wired together. The computational
device could include a laptop or a handheld computer, which collected and
stored the data for later download.
Maintenance for this would require downloading data but from one device,
and replenishing the power, but this time, for three devices.
·
Wearable Concept 3 –
data storage device wirelessly connected to two collectors. This concept is identical to Concept 2 but
the storage device is connected via wireless short-range communications, such
as Bluetooth, to the peripheral accelerometer and GPS device. Again, like Concept 2, this device would
have a download requirement from one device and a power-replenishing
requirement for three devices.
·
Wearable Concept 4 –
data conveyor wirelessly connected to two collectors. This concept is similar to Concept 3 in that there are no wires
connecting to any of the local (wearable) devices, but the storage device is
swapped out for a device which first collects the local data and then conveys
the data elsewhere, presumably, to a server maintained by the researchers. In this concept, the GPS and accelerometer
peripherals are wirelessly connected to a cell phone that transmits its
information through the internet to a server for storage. Here too, three
devices would require power replenishment.
However, data download is conducted automatically. Minus the accelerometer, this concept is
similar to that defined by Doherty, et al.[5]
·
Wearable Concept 5 –
data conveyor wirelessly connected to two collectors. This concept is identical to Concept 4 but a WiFi-enabled
handheld computer replaces the cell phone.
Data is collected from the peripheral devices and logged until the
central device locates a ‘friendly’ WiFi service, automatically connects, and
sends its information to a server. Like
Concept 4, power replenishment is the only frequent maintenance requirement,
unless a friendly network cannot be located in some amount of time.
·
Wearable Concept 6 –
integrated conveyor and one collector.
Here, the peripheral GPS device is replaced by a cell phone (conveyor)
with integrated GPS chip and wirelessly connected to an accelerometer. In this instance, there is no data
maintenance required; there is only power replenishment for two devices.
·
Desktop Concept – data
collection and sharing for analysis – The original concept for the database
design was to create an integrated system that is easy to upload information
into and display the results. The
components to this system included: a high-speed internet connection; a
dedicated IP address; a web server; a database; an automated mechanism to
upload data from field devices to the database; an automated mechanism to
graphically display location information; and a connection to the internet to
share collected data.
Refining concepts and evaluating technologies
Reinterpreting concepts into
reality, even such as those so well documented in the papers described above,
are often challenged by reality. Our
effort contended with the following requirements: test several technological
aspects of several concepts (size, weight, ability to provide useful data); a
5000-dollar technology budget; human resource restrictions that would allow our
study to mimic that which would be available to a likely future champion of
such a study: local governments and city planners. The considerations for technology selection in our effort follow.
Wearable Concept 1 called for
a wearable data-logging accelerometer and GPS device. The GPS device used by Oliveira, et al., was no longer available
at the time of this effort.[6] Further, the GPS device they had used
required the use of a backpack. Though
much less substantial than those used 5 years prior to the writing of their paper,
their GPS device had a small antenna placed on the shoulder pad of the backpack
of the participant and the primary electronics in a small pouch that would
reside within the backpack. Compared to
the scale of the accelerometer they were using, this seemed quite large. The GPS device they had used conflicted with
our effort’s focus on assuring minimal technological interference with the
individual wearer’s daily routine. Thus,
a low-cost, lightweight, and compact GPS data logger was sought out. Though GPS accuracy, data-storage capacity,
and battery life likely would suffer as compared with the GeoStats device, our
effort chose the Trackstick for its simple design and small size to serve as
GPS data logger in Wearable Concept 1.
As for the data-logging
accelerometer identified in Wearable Concept 1, the GT1M ActiGraph
accelerometer used by Oliveira, et al., was still available and affordable and
so was acquired for our effort.[7]
Wearable Concept 2 called for
a mouse-like accelerometer and mouse-like GPS to be simultaneously attached to
another device that could be used to store the collected data. This design quickly proved infeasible. First, two devices connected to another with
wires would become quite a hindrance to any wearer of the technology. Second, most mobile systems capable of
hosting two wired Universal Serial Bus (USB) connections (the preferred wired
connector for small, mouse-like, devices) are laptops, so the weight and the
cost would quickly skyrocket. Wearable
Concept 2 was no longer considered.
Wearable Concept 3 called for
two small, nearly unobtrusive devices, connected wirelessly to a third small
device that collects data and stores it for later download. Concept 3’s basic wearable wireless trio
design is consistent through Concepts 4 and 5 where two data providers are
sending data to a third data handler.
As such, technological acquisition could focus on a few technologies
that could double or triple time across the evaluation of each concept. Wearable Concepts 3, 4, and 5 all seek a
Bluetooth (local wireless) enabled GPS device.
This device would need to be fairly small, durable, reliable, accurate,
and easy to maintain (simple and infrequent charging). After exhaustive research as to the accuracy
of the newest available retail GPS chips, three GPS devices were selected.[8] Three different providers were selected
because as while the chips remained the same, the batteries, charging
technique, provided software, and the housing all varied somewhat. These were the Globalsat BT338; Sysonchip
Smart Blue Mini; and Holux GPSlim236.
Wearable Concepts 3, 4, 5,
and 6 all seek a Bluetooth-enabled accelerometer. At the time of the technology selection and acquisition phase of
this effort, there were very few options for this component. Through conversation with Mr. Doherty,[9]
a Bluetooth-enabled accelerometer was identified as being in production and
sale through the Australian firm, Alivetec.[10] Indeed, this firm’s product offering
included a Bluetooth-enabled accelerometer coupled with a heart monitor that
communicated its information to a Bluetooth-enabled cell phone and on to a
centralized monitoring service. This
concept is very much akin to those presented here. However, being coupled, the device had a cost of around $1000,
which was prohibitive in this effort.[11] At the time, the only other alternative was
to assemble a device based on the schematics outlined by a Georgia Tech
engineer.[12] While plausibly inexpensive in terms of
materials, is also seemed plausible that it could become quite expensive in the
time involved to identify a VTTI engineer capable of assembling the device, the
time to build the device, the time to calibrate and test the device in a way
that is behaved in a fashion similar to the device used by Oliveira, et al,
etc.[13] Thus, it was decided that the Bluetooth
accelerometer in the wearable concepts would be temporarily set aside and our
effort would use the ActiGraph device to serve as a fill-in as its data
captured and size would likely mimic a real Bluetooth accelerometer. Not surprisingly, by the time of the writing
of this report, there are more Bluetooth accelerometers available on the retail
markets. However, they are generally
tied in to complete, individual, health-monitoring packages, and as such, remain
relatively pricey, and their analytical software has been developed for a very
specific use.
Wearable Concepts 3 and 5
both call for their third device to be capable of storing the data for later
transmission. Where Concept 3 assumes
that the data transmission will be through a physical connection, Concept 5
augments that with a capability of transmitting the data via a ‘friendly’ WiFi
connection. That is, when the storing
device enters the range of a previously identified wireless network, and it determines
that there is an internet connection through that friendly network, it then
sends the collected data. There are a
lot of similarities between the two devices.
Concepts 3 and 5 both must have a Bluetooth capability and must be able
to store data. At the time of the
selection process for this effort, there were several capable technologies that
were small enough and relatively inexpensive enough to perform these basic
functions, not to mention capable of being programmed and connecting to a WiFi
service. Since a device with WiFi can
also store data, two devices were required to serve as a test device for two
concepts. Handheld PCs and Palm devices
were considered and of the wide variety of those available at the time, the
Hewlett-Packard iPAQ hx2495 Pocket PC was selected.
Wearable Concept 4 differs
from 3 and 5 in that the central collection device is designed to immediately
convey the data, through cellular service, through the internet, to a
centralized database. Where the selection
of the Bluetooth accelerometer was daunting in its lack of availability, the
selection of a cellular phone and service were daunting as our effort had to
consider a multitude factors: 1) identify the services available in Blacksburg,
Virginia and that would allow for functionality throughout the remainder of the
Commonwealth of Virginia; 2) identify a cellular-service provider that worked
efficiently with the State’s purchasing system (timely acquisition was a
consideration); 3) identifying a cell phone that would work with the service
and our devices. This last
consideration was particularly complex.
The phone would have to have a Bluetooth capability; it would have to
have to be programmable (preferably Java); and, since Wearable Concept 6 seeks
a cell phone with GPS to replace the external GPS device, the phone would also
have to have a GPS device that was integrated and accessible (to
programming). While the e911 mandate
has influenced cell-phone manufacturers to produce cell phones with integrated
GPS capability, there are few services that actually allow the owner of the
phone to access the GPS chip. As it
turned out, the company at the top of our list for service and phones provided
us with two technology demonstrators: Nextel provided the Nextel Motorola i605
and the Nextel RIM Blackberry 7100i as well as the necessary services to
facilitate our effort. In addition,
Nextel has a rich line of enterprise applications designed for businesses to
track their employees’ cell phones and let employees share their location with
fellow employees. As such, our effort
had the opportunity to use their services within our research context.
One further note on
technology availability: at the time of the writing of this report,
Bluetooth-enabled devices with WiFi capability, cellular capability, and
storage capabilities were just starting to become available. These would have been ideal to test; and
when they become available with GPS, they may too serve the purpose of
supporting Wearable Concept 6.
The Desktop concept will be
addressed further below. In essence,
however, it included a Dell Latitude running Microsoft XP and including Office
XP. It was connected to the internet
through a high-speed connection with a dedicated IP address. Any additional software installed on the
machine was derived from the specific technologies our effort acquired, such as
data extraction software for the accelerometer. Any additional software development, from the database to the web
services to the automated uploading to the programming of the disparate
wearable devices, utilized open source applications and forums. The choice for
this was simple: as inexpensive yet workable as possible.
The individual technologies
evaluated were either field (wearable) or desktop (analytical) in classification. Details on those acquired, configured,
programmed, and evaluated in this effort follow.
Field technologies
Nextel Motorola i605[14]
Product description – Nextel
Motorola i605 is a multifunction and rugged (US Milspec 810F) cell phone. The device used in this effort had an
integrated GPS location capability, was Bluetooth enabled, was Java
programmable, and could access the internet via the Nextel GSM cellular
network. The device is depicted at right
in Figure 1.
Intended use – Wearable Concepts 4 and 6 called for such a
device. In Concept 4, the device would
collect data via Bluetooth from both an accelerometer and a GPS device and then
retransmit it to a central server for storage and analysis. In Concept 6, the device would perform the
same function but would transmit its own GPS signal in lieu of an external
Bluetooth-enabled GPS device.
Actual use – Due to the inability to program this device as data
transfer conduits, only a hybrid of Wearable Concept 6, and not Wearable
Concept 4, was able to be supported by this device.
Cost(s) – $129.99 (i605) +
$32.79/mo. (Nextel Free Incoming 300) + $ 21.99/mo. (TeleNav Track Premium) +
$3.00/mo. (Public I.P.) + $10.00/mo. (Data Access) = $197.77 (for the first
month) and $67.78 (for each month following)[15]
General comments – The
strengths for this device are its ability to acquire, hold, and retransmit a
GPS signal as well as its overall ruggedness.
While the i605 is not as good as the stand-alone Bluetooth-enabled GPS
devices examined in this effort, in particular in terms of spatial accuracy,
signal acquisition and hold, it still performed reasonably well. The performance was good enough that one
could still develop a very good picture of human movement through a regional
space. The immediate drawbacks are its
aesthetics, size, and weight. Its bulk
could interfere with the daily routine of an individual, especially if there
were other peripheral technologies involved along with using this as a central
communications device. In addition, battery
life was less than anticipated.
However, its battery performed substantially better than the TrackStick
and the Blackberry 7100i – lasting at least a day before requiring recharge.
Nextel RIM Blackberry 7100i[16]
Product description – The Nextel RIM Blackberry 7100i
is a multifunction device that serves as a personal digital assistant (PDA) and
a cell phone. The device used in this
effort had an integrated GPS location capability, was Bluetooth enabled, Java
programmable, and could access the internet via the Nextel GSM cellular
network. The device is depicted at
right in Figure 2.
Intended use – Wearable
Concepts 4 and 6 called for such a device.
In Concept 4, the device would collect data via Bluetooth from both an
accelerometer and a GPS device and then retransmit it to a central server for
storage and analysis. In Concept 6, the
device would perform the same function but would transmit its own GPS signal in
lieu of an external Bluetooth-enabled GPS device.
Actual use – Due to the inability
to program this device as data transfer conduits, only a hybrid of Wearable
Concept 6, and not Wearable Concept 4, was able to be supported by this device.
Cost(s) – Cost(s) – $199.99
(i7100i) + $65.59/mo. (BlackBerry National Team Share 400) + $ 21.99/mo.
(TeleNav Track Premium) + $3.00/mo. (Public I.P.) + $10.00/mo. (Data Access) =
$300.57 (for the first month) and $100.58 (for each month following)[17]
General comments – The
Blackberry 7100i is an attractive device with many capabilities that were not
intentionally evaluated by our effort.
Such attributes of aesthetics and personal functionality might prove
useful in a future design where the central device being used to collect
information is also intended to be a benefit to the participant. The device, while technologically
fascinating, fell short of expectations on battery life (including the i605 and
the iPAQ, this device had the shortest battery life when transmitting GPS data)
and on apparent durability (it felt as if it needed to be cared for – seemingly
low-grade plastics). However, when it
was shielded in its holster and affixed to a belt clip, its relative flatness
allowed the wearer to go about activities without noticing. Of course, if the wearer were wearing
clothes without the need of a belt, placement might become more challenging –
though, its performance (GPS acquisition and transmission) suggested that
carrying it in a purse or a backpack would not be unreasonable. The caveat was that this device seemed to
have a tougher time acquiring and holding a GPS signal than the i605.
Hewlett-Packard iPAQ hx2495 Pocket PC[18]
Product description – This device is classified as a
handheld, or pocket, PC. While
diminished in performance, it has the capability to perform most tasks that a
laptop or desktop PC using conventional operating systems. Ours was configured with an additional 2
GigaByte (GB) Secure Digital (SD) memory card for additional storage. The device had an additional Compact Flash
(CF) slot for additional hardware peripherals.
This iPAQ also had Bluetooth and WiFi communications (802.11b)
capabilities. The device is depicted at
right in Figure 3.
Intended use – Wearable
Concepts 3 and 5 called for a third device to store and send collected data
respectively.
Actual use – Due to a
communications problem with this device and the local ‘friendly’ network (a
Virtual Private Network (VPN) connection was unable to be established and
therefore automation of the data collection and dissemination process was not
feasible), Wearable Concept 5 could not be evaluated. Therefore, this device served as the storage device for Wearable
Concept 3 only.
Cost(s) – $419.99 (iPAQ) +
$69.65 (2GB SD card) = $489.64
General comments – The
GPS/PDA combination worked well with a messenger-style bag with the GPS unit
attached to the shoulder strap and the PDA inside the bag. The PDA is a bit large/heavy/awkward when
carried in a pants pocket. There have
also been a few instances where the power button on the PDA was unintentionally
pressed in the both the bag and pocket, this problem was corrected by moving
the PDA into its own section within the bag and removing anything else from the
pants pocket.
The device did require a
great deal of configuration just to achieve Wearable Concept 3’s design
requirements. First, the device needed
to be configured to communicate with a Bluetooth-enabled GPS; second, and more
challenging, this device needed to be configured to store that information in a
data logging fashion; third, the device would need to be connected to a host PC
to download the data. The first and
third steps were relatively easy to accomplish. However, the second task required a third-party application to be
installed on this device. For this
project the data logger developed by KRMicros was used to record the data.[19]
The iPAQ was not designed as
a ruggedized device. As such, the user
must treat it like the fragile computer it is; it is not water resistant and it
does do well with temperature extremes.
Addressing such flaws, as well as making it a smaller, would make this a
very attractive candidate for field use.
GT1M ActiGraph[20]
Product description – This device is a wearable
accelerometer and data logger. An
accelerometer detects shifts in acceleration.
In this instance, the device can be used to measure human motion in one
dimension. Thus, when affixed to an
individual’s waistline, (e.g., belt) the device can measure acceleration along
that plane. In such a capacity it can
capture information on an individual’s gait and serve as a kind of pedometer
(measuring step frequency). The device
is depicted at right in Figure 4.
Intended use – Wearable
Concept 1 called for a data-logging accelerometer to serve as either a
pedometer or a device to measure gait characteristics. The GT1M served that purpose.
Actual use – Due to the
inability to acquire a Bluetooth-enabled accelerometer, the ActiGraph would
serve Wearable Concepts 3 through 6 as a defacto Bluetooth-enabled
accelerometer. If the USB connection on
the GT1M were replaced with a Bluetooth connection, this seemed a quite
plausible design possibility.
Cost(s) – $399 (device) + $3
(connector cable) + $300 (ActiLife software) = $702 (it should be noted that
one could have more than one device associated with one license/instance of the
software)
General comments – In terms
of the hardware and its ability to collect data, our effort was very pleased
with the GT1M. It was small, rugged,
water resistant, and lightweight. It
was so convenient to carry on a belt that our testers frequently forgot they
were wearing the device. This has
tremendous benefits when conducting such a survey as it removes impedance from
daily activity. However, clothes that do not require a belt might prove problematic.
The device was also impressive
when it came to its ability to remain ‘active’ – that is, when recording data
at its most frequent setting, the device had memory capacity and battery
capacity to continually operate beyond a 24-hour period. This was better than any other device our
group tested. It was also fairly easy
to connect the device to a USB port for data download and recharging.
The device did have its
drawbacks, specifically the software that came with the device. Establishing the software connection between
the host computer and the GT1M was often troublesome. Configuring the device once a connection had been established was
also undependable; our group had a great deal of trouble with the time-to-start
feature.
Bluetooth GPS Units: (a) Globalsat BT338[21], (b) SYSONCHIP SMART BLUE MINI[22], (c) Holux GPSlim236[23]
Product description – Each of
these devices are:
·
GPS receivers using the
SIRF Start III chip;
·
Capable of transmitting
GPS information via Bluetooth technology;
·
Stand-alone devices that
contain an internal, rechargeable, battery.
Intended use – Wearable Concepts 3, 4, and 5 all
called for the use of a Bluetooth-enabled and independent GPS device. Though markedly similar, and containing the
same GPS chip (which had been selected as being the most accurate retail GPS
chip), the external encasement, the battery and its life, were expected to vary
somewhat. The Globalstat BT338 is
depicted upper right in Figure 5. The
SYSONCHIP SMART BLUE MINI is depicted middle right in Figure 6. The Holuz GPSlim236 is depicted lower right
in Figure 7.
Actual use – Due to the
inability in the allotted time of this effort to successfully program the
phones to connect to the Bluetooth devices and retransmit their data to a
central server, these devices were only used in the case of Wearable Concepts 3
and 5 where they were paired with the HP iPAQ.
Cost(s) – (a) $121, (b) $166,
(c) $108
General comments – These were very impressive
devices. They exceeded our expectations
on their ability to acquire and hold a GPS signal where other devices have
historically failed. When used on our
field trips the device in a car, it could be left on the dash or in a cup
holder, on the person, strapped on a belt (using rubber bands), strapped onto the
shoulder of a backpack (using rubber bands), left in an outer pouch in a
backpack, or placed in a pocket. All
this worked as long as the individual was outside and in fair weather. If inside, and in an outer room with large
windows, the device seemed capable of acquiring GPS signals.
Packaging was adequate for placement on car
dashboards, or in exterior backpack pockets in fair weather. However, it was not adequate for direct
exposure to water.
As to device maintenance,
plugging the devices in for power recharging was simple enough. Collecting data from them to the data
collection device was easy as well, and so long as the data collection device
had a contemporary operating system and had Bluetooth, establishing a
connection was fairly easy.
TrackStick GPS Data Logger[24]
Product description – The
TrackStick GPS Data Logger is a device that logs a GPS signal at some
predetermined rate. The device is
depicted at right in Figure 8.
Intended use – This device
was intended for use in Wearable Concept 1 as a wearable companion to a data-logging
accelerometer.
Actual use – The TrackStick
was used as intended in Wearable Concept 1.
Cost(s) – $258
General comments – The TrackStick, out of the box,
held appeal because of its simple, compact, and seemingly convenient design; it
looked as if users could immediately put it in their pockets, attached it to
their belts, or sling it in their backpacks.
Real-world use, however, revealed that the device had many shortcomings:
GPS signal acquisition took a great deal of time; enabling the device to
acquire a GPS signal took a great deal more effort than the other GPS devices
(orient the device toward the sky and wait 15 minutes); positioning the device
for the GPS signal hold required the device to be constantly in clear view of
the sky – not an optimal feature when carrying or conveying the device; the
device’s rugged enclosure was not as rugged as it seemed; when collecting GPS
data at 1-second intervals the batteries would wear down in less than a day,
and while this is adequate for daily use, the device did not have the ability
to recharge internal batteries – changing the batteries required a good deal of
effort; when it was necessary to download the data, while the software
application was adequate, the device required the internal batteries to still
have some charge (the USB port was unable to provide sufficient current to
extract data).
These are significant
drawbacks for a device being used in study of the movement of pedestrians,
especially if the participant would need to maintain this device on a daily
basis along with other devices.
However, minor modifications to the device would make it very
attractive: 1) making the device rechargeable via the USB connection; 2)
improve the rugged character (make it water resistant); 3) replace the existing
second generation GPS chip with a third generation, such as the SIFR Star III.
Desktop technologies
Product description: The
central system was composed of more than one product. In general, however, it was a Dell laptop, Microsoft Windows XP
OS, Microsoft Office XP, and additional software applications that provided
functionality to the overall desktop design (such as a web server) or select
wearable technologies (such as wearable technology specific communications
software – to transfer data from one device to this system).
Intended use: The central
server would automatically collect all data from all wearable field
technologies, archive them, and make them accessible via a commonly used
internet device. More specifically, the
original concept for the database design was to create an integrated system
that is easy to upload information in to and display the results. The components to this system included: a
web server; a MySQL-driven database; a simple html form with a Perl backend to
upload the GPS text files to the web server and input the corresponding data
into a database; an html page incorporating the Google Maps API to present the
data generated in XML from the database.
Actual use: Those components
of Wearable Concepts 1 and 3 through 6 that functioned had their data
downloaded to this location, massaged (put into a common format), uploaded to a
database for storage, and made available via a web portal (not on Google Maps,
however).
Cost(s): Estimated cost of a
laptop (similar capability to the one our effort used): $2500.
Estimated cost of a high-speed internet connection with a dedicated IP
(slower yet similar capability to that of our effort): $100/month.
General comments – The entire
development process of the desktop component was a series of trials and
error. The original concept called for
a web server. For this project Apache
was chosen as the web server and MySQL as the database because of their freely
available distributions. There were
some issues in creating the proper file permission and configuring Apache,
MySQL, Perl, PHP, and Apache Tomcat.
Because of these difficulties a preconfigured bundle of Apache, MySQL,
Perl, and PHP called XAMPP was used along with an optional extension for Apache
Tomcat.
After the web server was
finally configured, the next step was to create an upload script to allow file
transfers from remote locations. The
first attempt at this was to create a Perl CGI script, but the files would not
transfer (this could have been a result of improper file permissions). The second attempt was with a PHP script
using PHP’s built in File Transfer Protocol (FTP) functionality. A connection
was established between the client machine and the web server host but the test
files were not transferred (this could have also been a result of improper file
permissions). The files were
successfully transferred from a client machine to the server using a standard
FTP program.
The next part in the database
creation was to establish a system of tables in MySQL to store the data. First, a user table was established to
generate a unique ID for each possible combination of person using the devices,
GPS, and intermediary device. Second, a
main table was created to store the following information: user name; GPS
device; intermediary device; unique user ID; date of data collection; time of
data collection; combination date/time ID; latitude; longitude; altitude; use
of an accelerometer; measurement from accelerometer.
Then, two tables were created
to act as “dump” locations for extracting data from the text files, one for the
GPS data and the other for the accelerometer data. A series of SQL scripts were created to load the data, make
necessary changes/conversions, and update the main table with the proper information. A few issues/errors did arise in this whole
process. The National Marine
Electronics Association standard (NEMA-0183) data stream recorded from the GPS
devices did not have any date information associated with it and the time
information is recorded in Coordinated Universal Time (UTC), whereas the other
devices are recorded the time information in Eastern Daylight Time (EDT). The NEMA-0183 data stream logs latitude and
longitude in decimal minutes, whereas some of the other devices log latitude and
longitude in decimal degrees.
Finally, to provide text file
dumps for the system, each wearable concept had its own process for downloading
(though, this was not the original intent).
·
Data collected from the
iPAQ and Bluetooth-enabled GPS (Hybrid Wearable Concept 3) were downloaded via
a direct connection made to the iPAQ through a USB sync cradle connected to the
central laptop. The file was a text
file in NEMA-0183 format.
·
Data collected from the
7100i and the i605 (Hybrid Wearable Concept 6) were done so through the Telenav
application portal on the internet (the latest version of Internet Explorer is
required, no other brand or version web browser will function). The downloaded files were in a MS Excel format
and could easily be converted into a text format. The downloaded information was not in NEMA-0183 format, but
rather only included: unique ID, time/date stamp, latitude, longitude, and
geocoded nearest roadway address.
·
Data collected from the
TrackStick GPS (Wearable Concept 1) required the use of the application that
was provided with the device. The
proprietary software allowed for an extraction of data in Microsoft Excel
format, which could easily be converted into text format. The downloaded information was not in
NEMA-0183 format, but rather included: unique ID, time/date stamp, elevation,
latitude, and longitude.
·
Data collected from the
GT1M ActiGraph accelerometer (Wearable Concept 1, Hybrid Wearable Concepts 3
and 6) required the use of the ActiLife application that was an additional
purchase beyond the initial hardware.
The software allowed the download of the data from the device into Excel
format and could easily be converted into text format.
Conclusions and Recommendations
The conversations with
international, national, regional, and local disparate discipline professionals
clearly showed an interest in collecting information on the way people move
through space. That is a more general statement than stating that there is
interest in the way pedestrians move through space – which was the impetuous
behind this effort. It turns out that
pedestrians are of interest, but only as just one of the many modes of movement
behavior that people assume. For example, transportation planners are
interested in where all people move; how they (the aggregate) get from point A
to point B; pedestrianism is just one way that people can get from point A to
point B. Health and nutrition experts
are interested in how much energy people expend in a given day, being a
pedestrian is just one of the energy activity expenditures. Biomotion experts are interested in the
movement of the anatomy of people through space, again, pedestrianism is but
one form the individual being studied might take. Certainly there is an opportunity to classify things as
pedestrianism, but the key here is this:
these researchers, engineers, and academics are interested in the
movement of people, not necessarily
things like cars. But the study of
people in the real-world (not in a lab), has historically been limited due to
technological limitations. This has
meant that there have been restrictions on the ability to study people in the
real world. In lieu of this capacity of
study, some disciplines turned to peripheral representations of human movement,
and to such an extent that these techniques of studies have become commonplace
and dominant. Conventional
transportation planning and engineering is automobile-centric. The ability for our cities to manage the
relationship between land use and transportation frequently relies heavily on
trip generation models that are associated with automobiles counts – not
people.
With the awakening in the
1990s of a systemic view of the world, “develop for yourself a holistic view of
some system,” transportation found a rational flaw in the study of the movement
of people. Not everyone was being
captured. Planes, trains, ships, and
automobiles were being counted, but the other things were, for the most part,
not. To compensate, suggestions were
made to modify existing deployed techniques and technologies used for counting
automobiles to count pedestrians and bicycles – to get the whole picture of
human movement from point A to point B.
This seems to be a half-hearted approach. All people know from experience, that when we walk, we often
follow a near limitless set of rules and opportunities; following the
automobile’s road network is not a primary rule. Thus, anything that is designed to capture the flow or movement
of people will need to be as flexible.
Over the past six years technologies
have evolved to such a point where the potential for disaggregate tracking of
people from point A to point B has become possible. This effort attempted to develop deployment concepts and review
select advanced technologies and techniques to determine what it would take to
feasibly study the movement of people and provide information to multiple
disciplines, but it proved to be challenging. This effort was only able to
scratch the surface of what it would take perform such a deployment.
There are two interrelated
components to this study: a technological concept and evaluation consideration
component, and a social organization (knowledge and needs) component.
Technology
This effort initiated by
reviewing a few very compelling wearable concepts. Transitioning to the real world left our researchers challenged
with the number of technical variables that had to be addressed. Indeed, some of the findings of this report
are cautionary notes.
·
Configuration and
programming of the desktop component, the WiFi device, and the cell phones took
an inordinate amount of time and yet did not yield the expected results. The time spent attempting to configure and
program these devices took as much time as the literature and the expert review
combined. More than anything, this was
due to the number of variables in the novel configurations our effort attempted
to review.
·
The purchasing of
technologies to meet the initial technology concepts proved challenging. The most significant disappointment was the
inability for our effort to acquire a Bluetooth-enabled accelerometer. Shifting the entire project forward into
time a mere 6 months may have rectified this as technologies would have become
available.
·
The number of variables
referenced in the bullet above can be found in the attempt our effort made to
capture the details of the devices as they related to the human wearer, to each
other, to the desktop, and intermediary components and systems. It seemed as if every time we looked at a
part of the system, we would find new variables to study.
These comments on hurdles go
to one cautionary note: when it comes to the number and type the technologies
involved, simplify: the fewer the number and type of devices, the fewer the
variables and challenges. Further, and
this will be addressed in the section below, the stakeholders involved in such
a project should also include a technical knowledge base in the use of specific
technologies (such as an accelerometer or GPS or Java).
Despite these hurdles, the effort
was able to capture some insight into some of the concepts. In Table 1 below, the initial wearable
concept is compared with the evaluated concept; costs and number of devices
worn and maintained (downloaded and recharged) are presented, as they proved
critical.
Table 1- A comparison of initial wearable concepts as
compared to evaluated concepts. *It
should be noted that the cost of one wearable system and does not include
desktop component costs.
|
Wearable Concept |
Initial |
As evaluated |
|
1.
Disparate data loggers |
Data-logging
accelerometer Data-logging
GPS |
GT1M
Actigraph TrackStick Devices
worn: 2 Device
data download: 2 Device
recharge: 2 Cost*:
$960 12
month total: $960 |
|
2.
Data storage device wired to two collectors |
Wired
accelerometer Wired
GPS device Wired
storage device |
Not evaluated: Too heavy, too large, too
expensive, and inconvenient for a potential wearer. |
|
3.
Data storage device wirelessly connected to two collectors |
Wireless
accelerometer Wireless
GPS device Wireless
storage device |
Modification: Wired
accelerometer: GT1M Actigraph Wireless
GPS devive: (a) Globalsat BT338 , (b) SYSONCHIP SMART BLUE MINI , (c) Holux
GPSlim236 Wireless
storage device: HP iPAQ Devices
worn: 3 Device
data download: 2 Device
recharge: 3 Cost*:
(a) $1313, (b) $1358, (c) $1300 12
month total: roughly $1300 |
|
4.
Data conveyor wirelessly connected to two collectors |
Wireless
accelerometer Wireless
GPS device Wireless
data conveyor (cell phone) |
Not evaluated: Unable to program phones to act as data conveyors. |
|
5.
Data conveyor wirelessly connected to two collectors |
Wireless
accelerometer Wireless
GPS device Wireless
data conveyor (“friendly” WiFi access) |
Not evaluated: Unable to program device to communicate with “friendly”
network. |
|
6.
Integrated conveyor and one collector |
Wireless
accelerometer Wireless
GPS enabled data conveyor (cell phone) |
Modification: Wired
accelerometer: GT1M Actigraph Wireless
data conveyor: 7100i and i605 Devices
worn: 2 Device
data download: 1 (for the wearer – the GT1M), 1 for the researcher – the cell
phone) Device
recharge: 2 Cost*:
7100i
mo. 1: $1003 7100i
mo. + 1: $101 i605
mo. 1: $900 i605
mo. + 1: $68 12
month total: 7100i:
$2109 i605: $1646 |
The above table does not
reflect the desktop component. Nor does
it reflect the researchers or the facilities in which the desktop or the
researchers are housed. There was an assumption that these things were already
on hand and funded.
Before using these
technologies this effort determined that the things worth evaluating were
concept cost, data accuracy, and component size and weight. After using these technologies, our
assessment focused on cost, consistency, and convenience. Cost remained of interest because the fees
associated with each device were still relatively significant. Studies performed by Virginia Tech on
physical activity focused on using a pedometer and a hand written log by the
participants.[25] The technology budget allotted $25 per
participant for each pedometer. Our
individual systems were substantially more expensive. However, tracking does eliminate some user and researcher
bias.
Data accuracy was out
because, for these products, there were other efforts that had evaluated
accuracy. The GT1M Actigraph has
demonstrated its accuracy and relevancy.[26] The Bluetooth GPS devices with the latest
generation of GPS chips have also been documented as being accurate, and for
the purposes of planners wanting to know where in a region someone is, they are
more than sufficient.[27] The cell-phone GPS devices have also
demonstrated their ability to provide accurate positional information.[28] The TrackStick data logger, however, had not
been tried in research papers. Thus,
what our effort might keep in mind was the ability to consistently provide data.
Size and weight were inherent
in our effort’s revised view of the technologies. Convenience subordinated size and weight. For the wearer/participant, convenience
meant: how many technologies do I have to juggle and how many devices do I have
to plug in and for what purpose (recharging goes to battery life; data download
goes to complexity of process)?
Convenience for the researcher was similar, but it was more attuned to
how easy was it to collect data and review it.
There are some important considerations
to keep in mind when reviewing our effort’s work:
The evaluated version of
Concept 3 likely proved the best in data collection. Interestingly, this design, specifically the arrangement of the
architecture as worn on our evaluators, was similar to that arranged in by
Oliveira, et al.[29] Our
informal comparison between cell-phone-GPS and Bluetooth-GPS devices revealed
that the Bluetooth GPS could provide more consistent GPS data – there were more
frequent updates (every second) and the signal was held even with some
obstructions (heavy foliage or indoors in exterior rooms). However, there were significant
drawbacks. This system is still
relatively inconvenient. The wearer
must juggle three devices. The
wearer/participant must charge three devices.
The wearer/participant must download data from one device. Such a routine might lead to a
wearer/participant not wearing the device because it becomes too much of a hassle. Cost for one wearable system for 12 months
would be roughly $1300.
The technologies in Concept
1, like Concept 6, were fairly easy to carry.
Just in terms of weight, however, this was the lightest
configuration. The convenience for
wearing the equipment, however, cannot make up for the maintenance
problems. The GT1M was fairly easy to
maintain – plugging it in to download also meant plugging it in to charge, one
act. The TrackStick, however, proved
problematic. Downloading the data was
one act, but replenishing its batteries required great effort relative to all
other systems evaluated. Further, its
inconsistency in GPS signal acquisition and collection made it almost
incomparable with the others. Cost for
one wearable system for 12 months would be $960.
With Concept 6, there were
only two devices to carry. When it came
to maintain the devices, the wearer need only plug in the GT1M once to charge
and download data, and plug in the phone to charge, the data would be accessed
virtually by the researchers. This
ability alone – virtually accessing the GPS data – was a tremendous
benefit. Not only could the researcher
access the data virtually, they could view and save the data. There were certainly drawbacks to the
phones, for example the relative bulk of the i605 and the relative poor battery
performance of the 7100i, but both had characteristics that seemed to balance
those traits out (the ruggedness of the i605 and the extensive user features of
the 7100i). Cost for the 7100i system
for 12 months would be $2109. Cost for
the i605 system for 12 months would be $1645.
A final note on
technology: for any technology
deployment that seeks to outfit participants with technologies, the focus
should be on:
·
Removing the burden from
the participants (as invisible from their daily routine as possible: fewer
things to plug in, fewer things to process);
·
Removing the burden from
the researchers (less set-up time, less time attending to participants, less
time attending to data massaging);
·
Removing the burden from
the sponsoring organization (lowering cost).
Organization: knowledge and needs
Any of the aforementioned
technological concepts is technically possible – they are all based on existing
technologies. They can become
technically possible and feasible given the appropriate alignment of a supporting
organization, virtual or otherwise: how much money or people power will one
throw at a problem?
Recommending the appropriate
organization for a deployment focusing on the outfitting of technologies to a
body of participants is not appropriate or possible from this effort’s vantage
point. However, given the experience
with this effort, there are a few comments that might prove useful for others
to understand who may be organizing for a similar effort.
First, the group of
stakeholders ought to also include individuals technically proficient in the
components of the system. Our effort
attempted to develop a system based on a premise of a small planning agency
going about such a deployment. We were
successful in identifying and communicating the disparate disciplines relative
to data needs, but we put too little effort in identifying and communicating
the disciplines that could configure and program individual components, such as
programming Java, or expertise in GPS or accelerometers.
In addition, there is a very
broad range of disciplines and organizations interested in the movement of
people, in particular, instrumenting people and studying the results from
real-world movement. For a new
organization – a new project – the broad interest is there (enough to start a
project), the technology is even there (given some minor modifications), but
there appears to be imbalance in fiscal allocation. A very minor example: a health-and-human-nutrition study on human
activity was given enough technology money for 10 pedometers – roughly $250. The budget for this effort was $5000. Admittedly, there is a difference between a
deployment evaluation and a real-world study, but this can be an indicator for
discipline sensitivity of appropriate allocation for technology.
Alternatives and opportunities
There are, of course,
alternatives that could be further explored.
Evaluated Wearable Concept 1
consisted of the TrackStick GPS data logger and the GT1M ActiGraph wearable
accelerometer. That system was the
lightest weight of all of the concepts.
It was, however, somewhat inconvenient: two things to download data
from, two things to recharge. Also, the
individual performance of the TrackStick was less than expected. In contrast, the GT1M’s wearable performance
was better than expected. Given the
bare bones nature of both devices, one wonders why both have not been
integrated and what are the challenges to ‘build your own device’?
Evaluated Wearable Concept 6
consisted of the GT1M and the i605 or 7100i.
While both were larger than the TrackStick, the i605 and 7100i
individually served as two devices.
Further, the i605 and 7100i facilitated an improved measure of
convenience for both the wearer and the researcher. This system was very near ideal except for the cost. Removing the 7100i from the system, with
only the i605 version of Concept 6, the cost becomes more reasonable. If the accelerometer had actually been
Bluetooth-enabled, and the phone had been successfully programmed, this system
would have assured that the wearer only charge both devices – the data captured
automatically, archived, and ready for analysis. Pressing further with such an evaluation is seen as a primary
activity beyond this effort: two easy-to-use and easy-to-wear devices. An alternative to that pursuit would be to
seek a modification to the Nextel data collection application (TeleNav Track)
to accept additional data streams from the phone and associate them with GPS
points collected. With such a system,
there would be no need for researchers to develop their own desktop components.
If one were to remove the
GT1M from Concept 6 leaving only the i605, could the GPS data collected from
the device provide sufficient information so as to satisfy various disciplines
interests? With the evaluation done for this effort, i605 reporting was, at a
minimum, at 1-minute intervals. This is
more than sufficient for educating city planners as to the movement of people
through a city. But would it suffice
for other disciplines (such as biomechanics)?
For example, replacing the accelerometer with the GPS to measure gait
would require a very high reporting rate.[30] Presently, the maximum rate of collection
with the i605, using an application this effort did not evaluate, is around 3
s. Apparently, any more frequent
reporting has a significant effect on the device’s battery life. Using such a device for gait analysis would
likely call into question many other things, such as the device’s specific
positioning on the person (belt or chest?); what happens if the person uses the
cell phone as a cell phone? Though
there are hurdles, such application deserves further exploration.
Another opportunity would be
to establish a relationship with a cell phone company to provide regional user
location data. This has already been
done but for automobiles.[31] However, it was not accomplished using the
GPS-enabled location techniques now offered by companies such as Nextel, as
such, individual tracks might not have been as accurate as they could have
been. With cell phone companies
providing tracking information, individual privacy is certainly a dominant
issue. As with the effort involving
automobile tracking (via cell phone), privacy could be maintained by removing
personal information and providing unique identifiers for each track. However, part of the benefit of participant
tracking goes to the categorization that occurs of the individual being
tracked: gender, age, ethnicity, profession, etc. To access such information, perhaps, once the relationship
between researchers and cell phone company had been well established, research
funds could be used to offer cell phone users a benefit if they participate in
a tracking study. A derivative of this
would be to first contact a major institution who maintains a large number of
such phones with a company such as Nextel.
Perhaps, for business purposes, phones could be tracked, and with the
consent of individual users, tracked beyond the context of work.
The goal of the project
sought to develop an understanding of what would be required to undertake a
wearable-technology pedestrian survey in order to establish the background
necessary for proceeding with a real world deployment. The goal-associated practical objectives
were to identify those who would be interested in monitoring pedestrians and
develop and review architectures of wearable of technologies. The first finding in pursuing those who were
involved and interested in the study of pedestrians was that of a shift from
focusing on pedestrians to human movement in general. This shift was better suited for two reasons: 1) it facilitated
the identification additional disciplines interested in the individual movement
of a person through space; 2) the technological differentiation between devices
that tracked humans in general and pedestrians specifically are merging;
tracking a person walking and transitioning to other modes had become easier
than originally anticipated when framing this project, therefore it was no
longer necessary to focus on one mode.
Following the shift in
terminology, the next finding revealed that those disciplines interested in
human movement through a region are quite diverse, ranging from public safety,
to biomotion analysis, to transportation engineering, to city planning, to
business operations, and to health and physical activity. While each had its own unique level of
detail it wished to capture relative to human motion through a given
environment, each would benefit from a deployment that could provide real-world
location and movement characteristics of an instrumented individual.
The wearable technologies
this effort developed and evaluated included six wearable architectures and
found that at least two, with some modification, would be worth further
exploration. These two systems, given
modifications, provide a balance in cost, data accuracy, data consistency, and
convenience (size, weight, device maintenance [number of things to download
data from; recharge], and number of things to carry).
Additional work is
recommended on exploring the costs and trade-offs associated with the
modifications of the two concepts.
After such work, a limited and trial deployment is recommended where
there are few fielded architectures, fielded for an extended period of time,
but inclusive of multiple disciplines.
This would test the architectures in their ability to handle extended
temporal and environmental stresses, and it would determine which disciplines
might benefit the most from an extended deployment.
Appendix A – Communications
Members of the research team
Stephanie Baker, Virginia
Tech Transportation Institute, steph@vtti.vt.edu
Marc Evans, Virginia Tech
Transportation Institute, mhevans@vt.edu
Eric Howard, Virginia Tech
Transportation Institute, ejhoward@vt.edu
Britton Lovell, Virginia Tech
Transportation Institute, blovell@vt.edu
Rewa Mariger, Virginia Tech
Transportation Institute, rmariger@vt.edu
Aaron Schroeder, Virginia
Tech Transportation Institute, aschroed@vt.edu
Individuals and organization that provided information
Ashwin Amanna, Virginia Tech
Transportation Institute
Craig Karlin, Actigraph
Sean T. Doherty, Laurier
University
Debbie Freed, Virginia Tech
Office of Transportation
J. Dan Brugh,
Blacksburg-Christiansburg-Montgomery County MPO
Jody Bickel, Virginia Tech,
Urban Affairs and Planning
David Clarke, Virginia
Department of Transportation, Christiansburg Residency
Bill Herbert, Virginia Tech,
Department of Human Nutrition, Foods & Exercise
Kathy Hosig, Virginia Tech,
Department of Human Nutrition, Foods & Exercise
Charlie Klauer, Virginia Tech
Transportation Institute
Cheryl W. Lynn, Virginia
Transportation Research Council
David Patton, Virginia
Department of Transportation, Bicycle and Pedestrian Coordinator
Sgt. Scott Poff, Virginia
Tech Public Safety
Lisa Schweitzer, Virginia
Tech, School of Public and International Affairs
Jean Wolf, Geostats
Appendix B - Technologies (product web sites)
Actigraph GT1M - http://www.theactigraph.com/
Globalsat BT338 - http://www.globalsat.com.tw/eng/index.htm
Hewlett-Packard iPAQ hx2495
Pocket PC -
http://www.shopping.hp.com/webapp/shopping/product_detail.do?storeName=storefronts&landing=handhelds&category=handhelds&orderflow=1&product_code=FA674B%23ABA&catLevel=1
Holux GPSlim236 - http://en.holux.com.cn/product/search.htm?filename=gpsreceiver_bluetooth_gpslim236_gg_ggtx.htm&target=gpsreceiver00&level=grandsonson
Nextel Motoroloa i605 -
http://idenphones.motorola.com/idenProducts/phonesHome.do?phones=605
Nextel RIM 7100i - http://www.blackberry.com/products/blackberry7100/blackberry7100i.shtml
SYSONCHIP SMART BLUE MINI - http://www.looket.com/
TrackStick GPS Data Logger - http://www.trackstick.com/index.html
Appendix C – Literature review paper