The surging wave of R&D into
smarter and smarter road vehicles is proceeding within an official
classification of levels of vehicle automation. However, it lacks a comparable
categorization of the roads on which they are to operate.
Reducing accident cost and fatalities is a driving force for robocars. |
Asking the right questions is
often the key to finding the right answer. How we think about something
determines what kind of answers we get. Robocars are still an amorphous concept
in a state of flux. Let’s think more about their context -- or more appropriately,
their contexts.
Focus on Vehicles
The US Government’s National Highway
and Transportation Safety Administration (NHTSA) in 2013 published a policy
framework defining five levels of vehicle automation:
0 None --
driver in full control
1 Function Specific -- automated braking,
cruise control, etc.
2 Combined Functions -- two or more driving
functions work synergistically, e.g. cruise control and lane centering
3 Limited Self-Driving -- driver cedes
controls under certain conditions
4 Full Self-Driving -- the robocar of our dreams and/or nightmares
This heuristic classification framework has become part of the
vocabulary of engineers and commercial entrepreneurs working on the many public
and private programs to design, prototype, test and implement robocars. It
helps them communicate the technical subtleties that often escape casual
participants in professional and public conversations. As such, it facilitates communication.
It is a useful intellectual tool that helps a society’s progress.
Enter the Context
Today we hear about “smart cars
in smart cities” still not knowing precisely what the second half of the term
means. What is a smart city? In 2013 MIT’s Technology Review with
electric vehicles in mind offered this:
“In a smart city, every electric vehicle must have access to a charging
station within its driving range.”
Since then we have witnessed remarkable
progress in robotic sensor and visioning systems. Public officials who deal
with the public realm are joining conversations with individuals and
corporations focused on individual vehicles. Cutting-edge thinkers are
exploring visions of a smart city where electronic markers dot the urbanscape.
Robocars service a metro station need not travel across town, go out to suburbs nor go cross country. |
Vehicles then can do more than watch out for
cars, kids and other obstacles on the running surfaces around them. They inform themselves by using an array of
local points that themselves can communicate back. For example, real-time data
on congestion can alter trip itineraries.
Stationary data emitters can also inform congestion managers. These can
be small-scale operations such as parking and circulation departments at large
university and medical complexes. Alternatively they can be large-scale, sophisticated
traffic control centers, such as the New York City region's Transcom,
Two-Dimensional Robocaricity
The performance and safety of robocars in
cities depend not just on the hardware and software within vehicles. They are
likewise affected by the extent, complexity and intelligence of the paths,
lanes, streets, arterials and interstate infrastructure over which they run.
Robocars without roads make no sense.
Why haven’t NHTSA or others categorized
the contexts in which robocars are to run? It’s one thing to operate robocars within
a private campus, where traffic can be tamed or even eliminated on some
segments of the campus and where security and maintenance resources are
available. College campuses tend to be pretty not because of intellectual
debates and high culture.
It’s a whole different undertaking to
run driverless vehicles from Albany to Albuquerque, Boston to Boulder, or
Syracuse to San Francisco.
A Matrix to Think Better
Trans.21 proposes the following
categories of Robocar Context:
0 Secured and managed campuses
1 Supervised activity centers with public
streets and multiple private owners
2 Urban neighborhoods and residential or mixed
use districts with firm boundaries (physical or policy) and populations of 5000
to 50,000 or more
3
An entire city - central or suburban -
but within the jurisdiction of a single legal entity - aka City Hall
4 A metropolitan area with populations of
0.5-15 or so million residents
5 Inter-regional - long-distance trips such as
mentioned above, extending from sea to shining sea.
Combining these two dimensions, we get
the following application matrix:
|
Context
Scale
|
0
|
1
|
2
|
3
|
4
|
5
|
Vehicle Smarts
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
1
|
|
|
A
|
|
|
|
|
2
|
|
|
|
|
|
|
|
3
|
|
B
|
|
|
|
|
|
4
|
|
|
|
|
|
C
|
|
In the above robocar-context
matrix, Cell A involves driven vehicles that
automatically brake in a major activity center -- such as a large shopping
district or an airport.
Cell
B is a much smarter robocar in which the driver opts over to automation
and can regain control operating within a
college campus. Neither A nor B seems
too daunting.
However the challenges in Cell C are significantly more complex and
challenging. This is large driverless fleets operating throughout a
metropolitan region. The time for
overcoming these problems doesn’t need to block serious discussion and
implementations in A and B.