February 2018 11
y Video cameras. These provide good detection of
objects and color, but often have poor night time
y IR cameras. These detects heat, giving extra detail --
especially at night.
y Radar. Short range radar can detect vehicles over a
narrow view for several hundred feet and can help
y LIDAR. Light detection and ranging. Radar with
infrared (IR) light. Uses IR laser to generate a 360
degree view over a short distance to aid in object
y Ultrasound. Ultrasound detectors operate a short
distance, but provide positive detection of objects.
In the coming years, the current ADAS will eventually
be complemented by the V2X communications capability,
providing even greater safety. Even though the most recent
mandate has been cancelled, we should see another
mandate in the future as V2X will be needed for self-driving
vehicles as well as for enhanced ADAS.
What Makes AVs Work is Software
It takes in all the sensor and communications
information, attempts to identify objects and a wide range
of driving scenarios, then makes decisions about braking,
accelerating, steering, or otherwise.
This “sensor fusion” software uses artificial intelligence
techniques including machine vision, machine learning
with neural networks, and deep learning. This software
requires massive data processing capability which currently
is being provided by multicore processors and/or graphics
processors from Intel, AMD, Nvidia, and a few others.
AVs are not ready yet, but manufacturers are frantically
developing them. Most estimates put availability out five
years or more. Surveys indicate that potential buyers would
not yet purchase an AV; a lot of consumers say they are
afraid of them. Yet, the push is on, and we are going to get
AVs sometime in the future.
For the elderly, handicapped, blind, or other
compromised drivers, AVs may be a real solution to their
transportation needs. For people who enjoy owning certain
cars and driving, AVs aren’t really a practical idea.
AVs are more of a transportation appliance and not a
source of image, skill, pride of ownership, and driving fun
represented by real cars. The NHTSA’s claim that AVs will
save lives has yet to be proven.
Some final thoughts from my perspective:
y ADAS work great! I have most of the basic features
in my new SUV and they do indeed add to safe driving
in today’s demanding, competitive, and brutal driving
environment. The backup camera and automatic braking
have saved me multiple times. The adaptive cruise control
makes highway driving easier.
y It takes time to learn the various ADAS features
and adapt to the warnings and responses. The goal of
manufacturers should be to maximize the benefits while
not overloading the driver with distracting warnings.
Human interface design is critical.
y Will V2X communications systems increase the
distractions with more and different warnings?
y Will drivers need training to learn the use of ADAS,
V2X, and other safety systems? Who will offer this?
y Will AVs actually save more lives? Or, when AVs are
mixed with human drivers, will there be more collisions
and dust-ups than ever?
y What do the insurance companies say about all this?
Who is to blame in an AV accident? The passenger, the
other passenger, a driver, the car manufacturer, or?
y Bottom line, I am all for ADAS and V2X systems. From
my experience, they will and do make us safer. It won’t
surprise me if AVs end up as a minor auto market niche.
y All the R&D related to AVs is positive as it will
accelerate improvements in the sensors, processors, and
software that can be used elsewhere. NV
THE LATEST IN NETWORKING AND WIRELESS TECHNOLOGIES
Table 1. Source: Society of Automotive Engineers
Level Degree of Automation
0 No automation. A human driver performs all tasks.
1 Driver performs all functions, but ADAS may provide partial control of steering, braking, and acceleration.
2 Partial automation. Automated systems control braking, steering, and acceleration, but driver is still “on call” to monitor conditions and take control if necessary.
3 Conditional automation. Automated driving systems handle all driving activities, but the driver must still be available to take control as required.
4 High automation. Automated driving systems perform all driving tasks. Driver may still control the vehicle if needed.
5 Full automation. No driver needed, but a driver may take over if required.