2014 Toyota Advanced Safety Seminar - Ken Koibuchi

As prepared for

Toyota Advanced Safety Seminar
Wednesday, Sept. 4, 2014
Ken Koibuchi, General Manager of Intelligent Vehicle Development, Toyota Motor Corporation

Good morning everyone.

As I believe you will see from my presentation today, I have the most-fun job in the company.

In fact, safety is fun…and I can prove it.

Today I will offer you a progress report

on what Toyota has been doing recently
and where we are planning to go in the future…
in the area of intelligent vehicle technology.

The Toyota Advanced Active Safety Research Vehicle

was unveiled at the Consumer Electronics Show in Las Vegas,
nearly two years ago.

That vehicle, capable of fully automated driving,

was developed mostly at the Toyota Research Institute North America

only a few miles from here
and was meant to offer a future reference point
for a broad active-safety strategy at Toyota.

Toyota believes that someday,

in specific traffic environments,
vehicles will be capable of full automation
but with drivers always in control.

As is the case with zero traffic fatalities,

full automation is a target difficult to achieve.

In striving for perfection,

safety advancements can be made at an quicker pace.

Our goal is to bring to market as soon as possible,

high-level driving assist systems
for the mass market.

One big question will be; will the masses be ready?

Toyota spends

about a million dollars an hour on R&D
because we take a broad approach
to find answers to challenging questions.

In the area of advanced safety,

we have been comprehensive in exploring all avenues
that will get us closer to zero traffic fatalities.

In fact, the new era in mobility…is already in motion.

In the course of researching and developing automated driving technology, we have succeeded in developing…

“Automated” active safety systems and
“connected” or Intelligent Transport Systems,
which are both high-level driving assist technologies.

Connected vehicle technologies

require either Vehicle-to-Vehicle or
Vehicle-to-Infrastructure communications.

Whereas automated driving technologies…

rely on processing information
solely from on board sensors, GPS and cameras.

At last year’s ITS World Congress in Tokyo,

we demonstrated an Automated Highway Driving Assist (AHDA) vehicle on public roads
equipped with new cooperative and adaptive cruise control
and lane tracing features.

The system is capable of assisting the driver with steering, braking and acceleration tasks

by maintaining an inter-vehicle wireless connection with the vehicles in close proximity.

This year, we will demonstrate

a similar functionality, with certain updates to match US road conditions .

but without being connected by vehicle to vehicle communications.

Instead, this year’s AHDA vehicle

relies solely on automated driving technologies
to navigate through downtown Detroit.

The aim of both systems is to develop

a predictable and trustworthy relationship
between the vehicle and the driver.

Not only does that mean

that the vehicle can take control from the driver in a critical situation…
it also means that the vehicle can alert the driver
that the driver should take control
of a function or maneuver…
that the vehicle has not been able to properly process.

This year’s system also uses a driver awareness feature

that scans eye movement and hands-on-the-wheel
and alerts the driver if it believes
that insufficient attention is being paid
to the road ahead.

In other words, a full-time backseat driver.

The AHDA Lexus GS

that we will demonstrate for you tomorrow in downtown Detroit
will cover a course of diverse conditions.

This newest edition of the AHDA concept series

is an example of our progress in Intelligent vehicles,
where high-level automation
has been refined and improved
and made more marketable.

Toyota plans to market technologies based on the AHDA advanced driving support system in the mid-decade.

Since the 2013 CES,

we have also been moving quickly
with advancements in the AASRV concept series.

What I am about to show-and-tell

is part of the fun stuff I mentioned in my opening.

A universal truth is that

Europe, Asia and North America are major markets
with wildly diverse traffic environments
that must be tailored—programmed…
into the awareness, processing and reaction functions
of automated driving technology.

The original AASRV was tailored to American traffic environments

with ample room and high speeds.

As research and refinement progressed,

we needed to shift to a traffic environment
that was less expansive,
with slower speeds
and much different signage and infrastructure.

In the video we will show you today

we have three vehicles, all three featuring
our current level of fully automated driving equipment.

The grey vehicle is being tested at its home base in the general Ann Arbor area

on specific public highways…
with open, high-speed sections
the predominant environment,
along with quite a bit of rural road
where signs and signals and stripes
might not be in abundance.

The white vehicle spends most of its days at our test facility near Mount Fuji…

where it is driven dawn to dusk
confronted with demanding, decision-making scenarios
such as blind hilltops and blind corners of merging traffic,
and light- into-darkness tunnels with pedestrians on narrow walks

These vehicles feature significantly enhanced software than our earlier AARSV

that allow them to read and identify objects
such as vehicles, bicycles and pedestrians in Japan’s crowded traffic environment
with greater accuracy and less variation.

The vehicles are designed to perform 3 main functions;

fully acknowledge the 360-degree surrounding situation.
Automatically take the safest route
Amend, or follow-up on driver error with display, alarm and/or automatic reaction.

To accomplish this, they must

verify 3 dimensional data via highly-accurate recognition technology, improved mapping,
and extract and properly identify
all vehicle types, bikes and pedestrians.

The red vehicle that you see in this video,

demonstrates a completely new and more demanding test environment in which to operate;
the crowded boulevards around the Imperial Palace in Tokyo.

This route was recently used

to demonstrate this technology to prime minister Abe.

Please note…that is not Prime Minister Abe behind the wheel.

Camera vehicles in front of and behind the red target vehicle

record how it navigates around parked busses,
road work and pedestrians in full automated mode.

What you also see is the data overlay

of what the vehicle is seeing,
and how that matches up
with what the camera vehicles are seeing.

Finally, there is the mapping overlay

that allows a viewing comparison using GPS, as well as on-board maps, LIDAR and cameras,

that gives us a more three dimensional view of the accuracy of this system.

In full-automated mode, this vehicle

understands and obeys basic safety rules
keeps in its lane, maintains a safe distance
and avoids a crash by anticipating predictable movements
of dissimilar objects.

A large bus will probably not suddenly lurch into your path

Whereas a motorcycle might suddenly zip past

The point of this, of course

is that to successfully navigate through such congested traffic environments,
the automated vehicle must gather as much “good data” as possible as quickly as possible…
from as many sources as possible…
steering, braking and accelerating within extremely close tolerances.

Simply put, the better the data “input”, the better the processing and the better the maneuver.

Even in conditions that are challenging for the onboard sensors, such as in the rain, the vehicle processes information from many sources and judges how to drive at an appropriate speed.

We’re getting better, but there is still much improvement necessary in the following four key areas.

The first is that we need to set realistic expectations