IEEE Technology and Society Magazine - December 2019 - 67

and therefore will rely on visible or audible cues to evacuate. People will respond to these cues in unique ways,
which will be specific to their experience and capabilities.
In a typical fire alarm, most occupants will be able to walk
to an exit, but what about an unaccompanied visuallyimpaired person, or a wheelchair user? Or if a person
becomes incapable of moving, due to physical injury or
psychological response? Design becomes far more complex when considering wayfinding for these sets of people,
and pedestrian dynamics provides the potential to investigate these scenarios ahead of time.
The most important element of the safety design paradigm is the data used to calibrate the models themselves. This data comes either from real-world events, or
experiments, and without high-quality data the models
will not accurately simulate hypothetical scenarios.
Waiting to learn from real-world emergencies is not a
particularly proactive attitude when considering safety,
which leads to the development of experiments to generate data. There is a trade-off between ecological validity and ethical viability when performing these
experiments, limiting how much information is available
from a controlled environment. Previous work (such as
[21]-[23]) provides us with some of the insight needed
to accurately model what happens during emergencies
such as fires, but more work is needed. For example, to
create a fully representative model for visually impaired
underground subway users, a researcher might investigate how visually impaired persons react when an unexpected fire alarm goes off in an unfamiliar, enclosed
setting. However, depending on the experimental construct, this might not be possible given ethical concerns. There is difficult balance to strike when
researching and modeling behavior in emergencies,
especially with vulnerable groups. Some of the required
work has been done (e.g., [24]), but, as before, there is a
lot of work yet to be done.

Natural Emergencies
Possibly the greatest potential benefit of pedestrian
dynamics is in designing for extreme emergencies.
Global warming is changing the landscape in several
ways. Wildfires are becoming more common, flooding is
a product of both higher sea levels and more extreme
rain events, and all these events have a high potential to
cause loss of life.
Is it possible to design infrastructure to accommodate people, and improve survival rates in these situations? Pedestrian dynamics is often thought to only
encompass movement of people, but it can be extended to include interactions between people and any type
of transport, including cars, trains, and planes. The
impact of these emergencies is disproportionately visited on poorer and vulnerable communities, often
DECEMBER 2019

∕

It is far better to understand how
actual users will behave, rather than
make assumptions that give rise to
a "rules-of-thumb" approach.
because their needs mean they cannot readily move
long distances. In the aftermath of Hurricane Katrina
(2005), some of the evacuation problems stemmed
from drivers being unable to enter the city on safety
grounds [25], an issue that would not be present with
the imminent (read: in the next twenty years) advent of
autonomous vehicles. With a fleet of these vehicles requisitioned for emergency service, government agencies
would be able to implement evacuation plans that will
help save people who might otherwise die.
There remain several issues that need to be solved,
including whether the government would be able to requisition private car fleets (e.g., Uber, Lyft), or the optimal
routing of vehicles given the location of evacuees and
the vehicles themselves. These issues present designers
with the further questions, and the impact of their decisions will be far more significant.
Evacuation planning for large-scale emergencies is
commonplace across the world, but usually only
improves after things have gone wrong (again, see Hurricane Katrina). Ongoing work specifically looks at these
types of questions (for example [26]) and continues to
develop the ability to model the movement of people on
a large scale.
This work incorporates demographic information as
well as all travel modes, combines evacuation methodologies, investigates worst case scenarios, and accurately understands the impact of different evacuation
policies. By using this approach, understanding individual behavioral responses, and utilizing available computational power, it is possible to improve the outcomes of
these large-scale natural disasters. Furthermore, real
time modeling represents a huge upside when considering quick-onset emergencies (flash flooding, for example), and advances are being made in this area.
This work is at the frontier of research and industry,
but still suffers from significant drawbacks. As yet it is
not possible to accurately combine microscale (individuals in a building) and macroscale (millions of city
residents) to understand any potential pinch points.
Real-time modeling suffers from similar issues, but also
requires a up-to-date input data, which is not always
readily available. If these obstacles were to be overcome, it is clear to see the potential benefits to society.

IEEE TECHNOLOGY AND SOCIETY MAGAZINE

IEEE Technology and Society Magazine - December 2019

Table of Contents for the Digital Edition of IEEE Technology and Society Magazine - December 2019