Transportation and mobility are key to quality of life, equity of opportunity, and economic well-being in urban environments. Yet for travelers with disabilities, navigation along city streets and access to public transportation present significant challenges and often impose severe limits on freedom of movement and the ability to accomplish normal daily activities. To explore the role of technology in addressing these challenges, the Federal Highway Administration established the Accessible Transportation Technologies Research Initiative (ATTRI).

One focus of the ATTRI program is the problem of safe intersection crossing. "For a pedestrian who is vision impaired, navigating intersections poses a particularly daunting task," says Mohammed Yousuf, the director of the Federal Transit Administration's Office of Infrastructure and Innovation, and formerly the program manager for ATTRI at FHWA.

Most frequently, the pedestrian must rely on the directional sounds of traffic to first orient him or herself to the intended crossing direction and then to determine when it is time to cross. These decisions can be facilitated in situations where other pedestrians are present or if the intersection is equipped with accessible pedestrian signals, which provide audible and vibrotactile cues. In the absence of such cues when crossing unfamiliar intersections, a common strategy for a pedestrian who has a vision impairment is to remain at the intersection for one or more cycles of the traffic signal phases in order to acclimate and understand traffic patterns sufficiently before making a move to cross. Once a decision is made to cross, the pedestrian generally moves as quickly as possible, using cues such as the crown of the road at the middle of the crossing to gauge progress. Despite his or her best efforts, the safety of a pedestrian with a vision impairment often depends on the alertness and accommodation of the drivers of oncoming vehicles.

For pedestrians who have a mobility impairment, the basic challenge is getting across the intersection in the time that is allocated by the traffic signal control system, and, again, safety often depends on the patience of the drivers of waiting vehicles as the traffic signal changes phases.

"For pedestrians with a combination of disabilities," says Yousuf, "the situation is even worse."

To help address these challenges, Carnegie Mellon University, with funding from ATTRI, developed PedPal, a mobile smartphone application that enables pedestrians to communicate directly with signalized intersections and to influence traffic control decisions to their advantage. PedPal combines emerging connected vehicle communication technology with a recently developed real-time, adaptive traffic signal control system to provide for a safer and more efficient intersection crossing experience for pedestrians with disabilities.

How PedPal Works

PedPal encodes its user's personal crossing constraints (such as travel speed) and conveys both desired crossing direction and required crossing time to the traffic signal system upon arrival at the intersection. The signal system, which is optimizing movement of all sensed traffic in real time, will in return ensure that the pedestrian gets the requested crossing duration whenever it initiates the next crossing phase in the pedestrian's direction. During crossing, PedPal and the traffic signal system communicate while PedPal monitors its user's progress, and, if progress is determined to be slower than expected, then the signal phase is dynamically extended by the signal system to accommodate this longer passage requirement.

If the user provides PedPal with pre-planned route information, the traffic signal control system can anticipate the pedestrian's arrival time at the intersection. By factoring this information into its determination of the upcoming signal timing plan, the system will coordinate a more efficient crossing opportunity that streamlines the user's overall crossing time.

System Design

To realize the functional behavior, PedPal takes advantage of recent technology advances in two areas: (1) vehicle-to-infrastructure (V2I) communication, which has produced both hardware devices such as dedicated short-range communication (DSRC) radios and V2I messaging standards, and (2) real-time adaptive signal control, which provides the ability to install personalized crossing constraints to dynamically extend the crossing time when necessary and to proactively factor pedestrian arrival times into the intersection's signal timing plan.

PedPal is currently running on an iPhone, selected primarily for the accessibility features that it provides. With respect to V2I communication, the app provides both DSRC and cellular options, the former achieved via Bluetooth® communication between the smartphone and an attachable DSRC "sleeve," and the latter through a cloud-based server connection to the intersection. In both modes, standardized DSRC map data and signal phase and timing message types are used respectively to communicate relevant geometric information about the intersection (such as the number of lanes, lane width, etc.) and current phase information from the intersection (such as how much green time is left in the current crossing phase) to the PedPal app. Likewise, standardized signal request messages and signal status messages are used by PedPal to communicate crossing intent and requirements to the intersection. The PedPal app can be used at any signalized intersection that is adaptively controlled, and can compensate for the absence of pedestrian signals at the intersection.

For real-time adaptive signal control, PedPal currently relies on the recently developed Surtrac system, which was initially piloted and deployed in areas of Pittsburgh, PA, and is now operational in about a dozen North American cities. Surtrac is designed specifically for optimization of urban traffic networks, where there are multiple competing dominant traffic flows that change throughout the day. Surtrac takes a totally decentralized approach to traffic control. Each intersection allocates its green time independently in real time, based on actual incoming vehicle flows, as seen through video or radar detection devices. Then, once an intersection system has generated its timing plan, it communicates projected outflows to neighboring intersections to increase their visibility of future incoming traffic.

Timing plans at each intersection are executed in "rolling horizon" fashion, and the planning cycle repeats every second. Reliance on decentralized intersection control ensures maximum real-time responsiveness to actual traffic conditions, while communication of projected outflows to downstream neighbors enables coordinated activity at the network level and creation of green waves when appropriate. The real-time intersection scheduling procedure is sensitive to travel mode information (such as passenger vehicle, bus, bicycle, pedestrian) and capable of multimodal optimization if sensors (and in this case the PedPal app) can provide mode information.

Crossing an Intersection with PedPal

Once a pedestrian launches PedPal, the app begins "listening" for specific messages broadcast from nearby smart signalized traffic intersections, and it continues to be active until explicitly turned off. While active, the app continues to cycle through the screens shown here.

Screen (a): The "No nearby intersection detected" message indicates that the device is currently between intersections and offline.

Screen (b): When within range of the messages being broadcasted by an upcoming intersection, the app uses this information to display possible crossing options to the user, along with context about when the crossing direction is going to change. Upon arrival at the intersection, the user indicates crossing intent by tapping on one of the presented options, which can be displayed visually or via audio voiceover.

Screen (c): When the pedestrian makes a selection, PedPal automatically sends a message to the intersection indicating both the crossing option selected and the time required, and the app display transitions to a screen that provides guidance relevant to the selected crossing phase. If the user has selected a future crossing phase, PedPal warns the user that it is not safe to cross. If voiceover has been enabled, this message is conveyed through audio as well as visually. When the crossing signal is close to changing to the desired crossing direction, PedPal starts to count down the seconds to alert the user to get ready to go.

Screen (d): When the intersection's crossing signal switches to the user's direction, the app announces that it is OK to cross. At this point, the app shows the amount of crossing time that has been allocated and begins a countdown of the time remaining until the next signal change. Once the pedestrian starts to cross, the app begins to track crossing progress.

Screen (e): While in either waiting or crossing mode, the user can request available information to assist in orienting to the crosswalk at the current corner.

Screen (f): The countdown timer continues when the popup for more information is dismissed.

The PedPal Mobile App

The PedPal app is the interface between the pedestrian and the traffic signal control system. The developers designed it to promote ease of use and follow universal design principles while taking full advantage of iPhone accessibility features to provide support for pedestrians with different types of disabilities. The interface emphasizes selection from tabs and lists of options rather than requiring textual inputs, and it provides multiple modalities to the user for presenting information and options for selection.

For pedestrians who are not vision impaired, the app's visual interface provides a straightforward basis for using the mobile app. Upon bringing up the app, the user has a choice of selecting one of two screens: "nearby," which provides the user with information and crossing options for the upcoming intersection, and "settings," which enables the user to customize the app's behavior for specific constraints and preferences. During travel and intersection crossing, the user will typically operate solely from the "nearby" screen.

Users who have vision impairments can enable the smartphone's native accessibility features–including voiceover, font resizing, and text zooming–and configure them to facilitate interaction. The current PedPal prototype is implemented for the iPhone (in large part because of the relative strength of its onboard accessibility features), but the developers anticipate Android implementation in the future. Finally, the PedPal app is also designed to exploit haptic (vibration-based) cues for announcing walk/don't walk crossing conditions and for issuing alerts during the act of crossing.

Field Experiments

Carnegie Mellon carried out initial field testing of the PedPal mobile app at multiple intersections in the existing Pittsburgh Surtrac deployment to assess the app's potential and to obtain feedback for additional development. The development team recruited 14 people from the local Pittsburgh disability community, spanning a range of demographics and disability types but including a disproportionate number of individuals who have vision impairments. The team first interviewed each participant to form an understanding of his or her current challenges when crossing traffic intersections. Then each participant was trained in the use of the PedPal app and asked to perform a number of crossing trials both with and without the assistance of the app.

The developers used the crossings without the app as a control study for comparison with each user's crossing performance when using the app. The team recorded several measurements for each crossing trial, including the amount of time the user waited upon arrival at the corner, the number of traffic signal cycles that the user waited through while preparing to cross, and the eventual crossing duration. After the trials, participants were asked to complete a survey designed to get their qualitative assessment of the technology and suggestions for improvement.

Overwhelmingly, the feedback from participants was positive. Although it was clear that the app was perceived to be of more benefit to individuals with specific types of disabilities (such as walker users and pedestrians with vision impairments), all participants were enthusiastic about PedPal's potential to increase safety and enhance mobility for pedestrians with disabilities. One participant observed, "I rely on my Seeing Eye® dog to safely guide me across ... street crossings, and my Seeing Eye dog relies on me to confidently give the command to initiate a crossing. The PedPal app gives me the information I need to make an intelligent decision at a crossing, eliminating the need to interpret the cacophony of vehicular cues, often the lack of vehicular cues, and the occasional misinformation from fellow pedestrians."

Dr. Tessa McCarthy, an orientation and mobility specialist at the University of Pittsburgh, was also enthusiastic. "The PedPal app is one of the most innovative tools for street crossings that I've seen in my career," she says. "It has the ability to solve a lot of problems related to street crossings for people with a variety of disabilities."

Quantitatively, the research team found that the use of PedPal reduced both the total wait time and the number of cycles needed to cross for participants who have vision impairments. However, these participants actually moved across the intersection at a slower, relaxed pace when using PedPal. Based on observations and post-trial discussions with participants, this change in pace appears to correlate to the confidence users feel when crossing the street with the app: they feel more reassured that they are crossing at the correct phase and are less frantic in their movements as a result.

The observed results also confirmed prior expectations about the time required for crossing by different demographic groups. Individuals with mobility challenges who are walking require the most crossing time. Individuals who have vision impairments also require extra crossing time during periods of high traffic congestion. Alternatively, motorized wheelchair users and people with hearing loss cross intersections efficiently and do not need extra crossing time.

Next Steps

While a technological solution such as the PedPal app may not be right for every pedestrian, it can help many people with disabilities. Carnegie Mellon is currently expanding the PedPal technology to provide additional safety and mobility-enhancing capabilities, including the ability to detect veering outside of the crosswalk, to provide real-time feedback in an accessible format to pedestrians who have vision impairments, to import and export route information from third-party navigation apps, and to broadcast the presence of pedestrians with disabilities at the intersection to approaching connected vehicles. Carnegie Mellon is working with the company responsible for commercializing the Surtrac traffic signal control system to make the PedPal app available to any local municipality that has deployed Surtrac in the near future.

"The PedPal project has taken an important step toward realization of ATTRI's vision of an accessible transportation network that provides safe, easy, and efficient traveling options to everyone, especially to people with disabilities," says Yousuf.

Govindarajan Vadakpat is a highway research engineer at FHWA, where he managed the ATTRI safe intersection crossing project. He holds a Ph.D. in civil engineering from Penn State.

Stephen F. Smith is a research professor of robotics at Carnegie Mellon University and principal investigator of the PedPal project. He holds a Ph.D. in computer science from the University of Pittsburgh.

Zachary B. Rubinstein is a principal project scientist at Carnegie Mellon University and project manager of the PedPal project. He holds a Ph.D. in computer science from the University of Massachusetts at Amherst.

M. Bernardine Dias is a recognized world leader in the application of technology for social equity, and consulted on the PedPal project. She holds a Ph.D. in robotics from Carnegie Mellon University.

For more information, visit https://rosap.ntl.bts.gov/view/dot/50548 or contact Govindarajan Vadakpat at g.vadakpat@dot.gov or 202–493–3283.