to the gap as “comparing climbing Mount Fuji to
climbing Mount Everest.”
A Two-Way Street
To be truly accurate, next-gen digital road maps
must be constantly updated to reflect changing
traffic conditions and patterns, weather conditions
and laws. HERE’s maps incorporate three kinds of
information, says Mr. Rabel: basic road information (mostly static), a dynamic “live roads” feed
(detailing traffic flow, accidents, lane closings) and
behavior analytics. The latter involves long-term
analyses of driver behavior on specific roads in
specific situations. “We are tapping into different sources, merging them together, harmonizing
them and then applying changes to the map,” says
Mr. Rabel.
It’s a two-way process. Driverless vehicles aren’t
just passive recipients of road information: They
constantly bolster the accuracy of maps and their
depth of detail by using cameras and other sensors
to gather data that validates or corrects the maps.
For example, a vehicle’s camera could recognize a
speed limit sign and compare the posted limit to
the one on the map; if there’s a discrepancy, the
vehicle could report a needed change. Or a vehicle’s
Mapping the Future
Maps have guided people around the world for
millennia. Now, at the dawn of the driverless age,
project teams are working to produce maps of
unprecedented accuracy that can guide vehicles to
safety. But they must first meet intense technical
requirements.
The viability of automated vehicles depends
on highly detailed digital maps updated to reflect
changing road conditions. There’s no room for
error: Being off by as little as a few meters or yards
could send a car onto the wrong side of the road.
Or off a cliff.
“When you create a map for an intelligent
car, you give the car instructions. You can’t give
instructions that say, ‘Turn right in 90 meters’
when in reality the road is in 100 meters,” says
Dietmar Rabel, director of product management in
the highly automated driving group at the digital
mapping firm HERE, Frankfurt, Germany. “It has
to be very precise and reliable.”
HERE is one of several organizations around
the world working to bridge the large gap in
quality between current GPS maps (which
are supplemented by drivers’ common sense)
and those for automated driving. The CEO of
another digital mapping firm, Zenrin, referred
HERE’s HD map captures billions
of 3-D points to model the road
surface. Top right, HERE maps the
streets of Berlin, Germany.
“When you
create a map for
an intelligent
car, you give the
car instructions.
It has to be very
precise and
reliable.”
—Dietmar Rabel, HERE,
Frankfurt, Germany
theEdge
IM
A
G
ES
C
OUR
T
ESY
OF
HE
R
E