25-06-2014, 10:35 AM
Using a Smartphone App to Assist the Visually Impaired at Signalized Intersections
Using a Smartphone App to Assist the.pdf (Size: 1.29 MB / Downloads: 21)
INTRODUCTION
Individuals who are blind or visually impaired use the auditory and limited visual information
that they can gather to make safe crossing decisions often report being dissatisfied with a general
lack of information while crossing intersections, as found by Ponchillia, Rak, Freeland, and
LaGrow (2007). This may explain why a study of blind pedestrian behavior in three cities found
that only 49% of the crossings started during the walk interval (Barlow, Bentzen, & Bond, 2005).
They also found that 27% of all crossings (that did not involve outside assistance) ended after the
onset of the perpendicular traffic stream.
At crossings where using a pushbutton was required, Barlow, et al. (2005) found that few (0% -
16% depending on the city sampled) looked for and found the button; they also began walking
only 20% of the time during the walk signal compared to 72% of the time when the pedestrian
phase was on recall (i.e., included in every cycle). This may be because searching for the button
often requires the pedestrian to move from their path of travel, which is often used as an
alignment cue to make sure they are crossing straight. This suggests that there is room for
improvement in terms of the design and accessibility of both accessible pedestrian signals (APS)
and non-APS crosswalk signals for blind and low-vision pedestrians.
Research Objectives
The objective of this project is to evaluate and validate a smartphone-based decision support
system while providing geometry and signal information to the blind and visually impaired
pedestrians at signalized intersections. Both objective and subjective measures were developed to
evaluate system usefulness, trust and user’s satisfaction.
Brief History of Accessible Pedestrian Signals (APS)
The audible pedestrian signals first appeared in 1920’s in U.S. However, it is not included in the
US standard, Manual on Uniform Traffic Control Devices (MUTCD) until 2000. In mid 1970’s,
the audible signals were mounted on top of the pedestrian signal display (also called pedhead-
Literature Review
It may be arguable that providing wayfinding technology for the blind and visually impaired may
undermine the maintenance of their learned techniques. However, the application to improve
safety and increase capability for the visually impaired is more likely to outweigh the overall
cost (Loomis et al., 2007). Navigation and wayfinding involve with dynamically monitoring a
person’s position and orientation with respect to the immediate environment and destination
(Klatzky et al., 1998, 1999; Aslan and Krüger, 2004; Rieser, 2007). Navigation usually implies
that a user will follow a predetermined route or path between a specific origin and destination.
Navigation is often referred to as an optimal path based on a specific goal, such as shortest time,
distance, minimum cost, etc. However, wayfinding refers to the process of finding a path, not
necessary traveled previously, between a pair of origin and destination. The wayfinding process
is more adventurous and exploratory.
Starting in 1990, the American Disability Act (ADA) requires built environment accessible to
people with disability (Bentzen, 2007). Blind people are more vulnerable to collision due to
insufficient information (such as distant landmarks, heading and self-velocity) and time for
planning detour around obstacle (Loomis et al., 2001, 2007). People with wayfinding difficulties,
such as visually impaired (Golledge et al., 1996; Helal et al., 2001), elderly people (Rogers et al.,
1998; Kirasic, 2002; Hess, 2005), dementia or Alzheimer's diseases (Uc et al., 2004; Rosenbaum
et al., 2005; Pai 2006), can benefit from a personal navigation system for navigation and
wayfinding assistance. There has been lots of research investigating Geographic Information
System (GIS) and Global Positioning System (GPS) based navigation system for visually
impaired pedestrian (Golledge, et al., 1991, 1996, 1998, 2004; Helal et al., 2001; Ponchillia et
al., 2007; Blake, 2011). Several researchers also focused on the development of User Inte
Blind Pedestrian at Intersection Crossing
People with vision impairment use auditory and limited visual information that they can gather to
make safe crossing decision at signal intersection. They generally have difficulty crossing
intersections due to the lack of information available to them about the traffic and geometry at
intersections (Ponchillia et al., 2007). A study of blind pedestrian's behavior in three cities found
that only 49% of the crossings started during the walk interval (Barlow et al., 2005). The study
also found that 27% of all crossings (that did not involve outside assistance) ended after the onset
of the perpendicular traffic stream.
At crossings where using a pushbutton is required, Barlow, et al. (2005) found that few (0% -
16%) looked for and found the button; they also began walking only 20% of the time during the
walk signal as compared to 72% of the time when the pedestrian phase was on recall. The reason
may be because searching for the button often requires the pedestrian to move away from their
path of travel, which is often used as an alignment cue for crossing. In addition, Barlow et al.
(2005) found that although 72% of blind participants started with appropriate alignment,
location, or both, 42% ended their crossing maneuver outside the crosswalk. Guth et al. (2007)
found that site-specific characteristics (for example, treatments such as rumble strips or speed
countermeasures) appeared to have a greater impact on reducing the number of conflicts between
pedestrians and vehicles than did a mobility device (e.g., cane or guide dog). Therefore,
enhancing pedestrians’ ability to perceive useful cues at an intersection may be an effective
Navigation Technology and Location Based Services (LBS) for the Blind
Development of travelling aids based on global positioning has a long history. The first satellite
navigation system, used by US Navy, was first tested in 1960. The use of GPS to guide blind,
visual impaired or elderly people has been researched extensively (Garaj, 2001; Gill, 1997; Helal
et al., 2001). Tjan et al. (2005) designed and implemented a Digital Sign System (DSS) based on
low-cost passive retro-reflective tags printed with specially designed patterns. Blind or visually
impaired pedestrians can use a handheld camera and machine-vision system to identify and
navigate through unfamiliar indoor environment. Bae et al. (2009) evaluated a location tracking
system using IEEE 802.11b Wi-Fi system to analyze the requirements of location based services
in an indoor environmen
Report Organization
The rest of this report is organized as follows. The mobile Accessible Pedestrian Signal (MAPS)
system architecture and software design are discussed in Section 2. Experiment design,
procedures and performance measures are included in Section 3. Experiment results are
presented in Section 4. Lesson learned and future works are discussed in Section 5. Finally,
conclusion is presented in Section 6
MOBILE ACCESSIBLE PEDESTRIAN SIGNAL (MAPS)
After receiving orientation and mobility (O&M) training from O&M specialists, people with
vision impairment usually can travel independently to known places along familiar routes by
relying on a white cane or guide dog. However, neither the white cane nor the guide dog
provides spatial awareness along a path or traffic signal information at an intersection. Travelers
with vision impairment may have difficulty planning or feel anxious about unfamiliar routes,
often obtaining and closely following detailed turn-by-turn instructions to reach new
destinations.
In addition to a cane and guide dog, many aids based on various types of technologies have been
developed in the past. Several of these are commercially available to support both indoor and
outdoor wayfinding and navigation for people who are blind. Several of these technologies aim
to address the indoor and outdoor wayfinding problem separately. These solutions generally do
not provide intersection geometry and traffic signal information for pedestrians who are blind or
visually impaired
System Objectives
The first objective is to provide intersection geometry, traffic signal information and automatic
pedestrian phase request through a smartphone device. A smartphone application prototype will
integrate available sensors on the phone to determine a user location and orientation with respect
to an intersection. The smartphone app then wirelessly communicates with a traffic controller
and receives near real-time signal timing and phasing updates. Corresponding signal phasing
information is sent to smartphone according to the desired direction of crossing as confirmed by
the user. After receiving confirmation from users, the system will then provide timing
information of corresponding pedestrian phase to users. Warning signals such as ‘Do not walk’
or ‘Walk phase is on, # sec left’ will be broadcasted through Text-to-Speech (TTS) interface to
support decision making at crossing. Automatic pedestrian pushbutton request can also be sent to
signal controller for registered blind, visually impaired, elderly or disabled pedestrians when they
confirm the desired direction of crossing.
System Architecture
The system diagram of MAPS data communication is illustrated in Figure 2-1. For field
experiment, a Smart-Signal unit, or traffic Data Collection Unit (DCU), was installed in a TS2
traffic signal controller cabinet to obtain signal timing and phasing information through the
Synchronous Data Link Control (SDLC) interface, i.e., Bus Interface Unit (BIU). Real-time
traffic phasing and timing information is wirelessly transmitted to a signal database server
located in Minnesota Traffic Observatory (MTO) in the Civil Engineering Department at
University of Minnesota. When users perform a single-tap on the smartphone screen while
pointing to a desired direction, the smartphone app will interact with the signal data server to
obtain geometry information. After determining which direction to cross, the visually impaired
pedestrians can simply perform a double-tap on the smartphone screen to submit pedestrian
crossing request. The double-tap action then sends the request to traffic controller cabinet
through the signal database server. The pedestrian call interface, residing in a Virtual Private
Network (VPN), handles the request from the smartphone to active the pushbutton inputs in the
controller cabinet. The VPN is to ensure system security that no unauthorized users or systems
can trigger the pushbutton request remotely. Ideally, the wireless router/modem will not be
necessary when the signal data can be accessed through the MnDOT's firewall in the
EXPERIMENT DESIGN
Documentation of proposed field experiment was submitted to and approved by University of
Minnesota Institutional Review Board (IRB Code # 1112S07962).
Materials
The experiment at intersection #1 took place nearby a commercial area on a light raining day
from 7:40 AM to 5:30PM. The experiment at intersection #2, two blocks away from the first
intersection, took place on a bright sunny day from 7:40 AM to 5:30PM. Sidewalk areas were
clear with no obstacles at both locations. However, the sidewalk surface, at couple spots, was a
bit uneven at intersection #2. A traffic data collection unit (DCU) and a relay IO module were
installed in the controller cabinet at intersection #2 to provide signal information to smartphone
users. Survey questionnaire regarding usability, acceptance, and trust of the smartphone-based
traffic signal system was read to each individual and their responses were recorded on the
questionnaire. See Appendix B for questionnaire details.
Procedures
At each location, participants were asked to perform crossing tasks and then interviewed by a
research team member before and after each crossing task. The crossing tasks and interviews
focus on participants’ experiences while crossing signalized intersections, using audible
pedestrian signals, or a smartphone-based accessible pedestrian signal device provided by the
research team.
For each crossing task, a certified O&M specialist brought each participant to a starting point
which was located about 100 to 200 feet (north) away from the north-east corner of the
intersection, as illustrated in Figure 3-1. Visually impaired participants were asked to travel
along the sidewalk using their own navigational skills to reach the corner of the intersection.
While at the intersection, the visually impaired participants need to find and use the pushbutton
to request a pedestrian walk signal or use the smartphone-based pedestrian signal device to
determine when it is possible to cross. Participants then cross the street that is perpendicular to
the sidewalk they just travelled and arrive at the other side of street.
SUMMARY AND CONCLUSION
People with vision impairment generally have difficulty crossing intersections due to lack of
information available to them about the traffic, signal and intersection geometry. Among the
intersection crossing sub-tasks, locating the crosswalk, determining when to cross and
maintaining alignment with the crosswalk while crossing are the most difficult tasks for the blind
or visually impaired to execute. The current Accessible Pedestrian Signal (APS) system requires
the blind to search for a pushbutton if one even exists. It often requires the pedestrian to move
away from their path of travel, which is often used as an alignment cue for crossing. Due to the
high cost of the APS installation, most agencies do not deploy them at all signalized
intersections. In addition to the installation and maintenance costs that accrue to the local traffic
agency, current APS systems contribute significant “noise” to the local neighborhood.
Furthermore, the auditory guiding cues provided by the APS are often inaudible because of the
ambient traffic noise associated with rush hour. There is room for improvement in terms of the
design and accessibility of both APS and non-APS crosswalk signals for blind and low-vision
pedestrians.
Among the intersection crossing sub-tasks, locating crosswalk, determining when to cross and
maintaining alignment to crosswalk while crossing are the most difficult tasks for the blind and
visually impaired. We have interviewed ten blind and low-vision people to understand what
types of information they use at intersection crossings and identified information types that could
assist them. Six high-level recommendations emerged for the design of MAPS: