Page 13 - EE Times Europe Magazine – June 2024

EE|Times EUROPE 13

The Autonomous Car Industry in 2024: Sensors, Software and Safety

Thanks to improvements
in-car connectivity, onboard
hardware capabilities, and
a global shift in consumer
preferences toward subscrip-
tion models, the SDV market
is set to grow at 34% CAGR
between 2023 and 2034.
(Source: IDTechEx)

such information, a connected car could react driving without a single disengagement. By they be better than every human driver? But
appropriately to the location and movement comparison, IDTechEx estimates that human how would that be defined? A 17-year-old
of every pedestrian, cyclist and vehicle on the drivers in the U.S. are involved in collisions with a fresh license may never have crashed,
road around it, including those beyond its approximately once per 200,000 miles, and in i.e., one collision per infinity miles.
line of sight. Combine collective perception cities like San Francisco, this rate nearly dou- Complicating matters is that the true safety
with a responsible autonomous driver, and a bles. So perhaps the argument can be made record for human driving is not fully known.
country full of such cars could end road traffic that autonomous cars are already safer than Not all collisions are reported, and not all
accidents altogether. human drivers. dangerous situations or near-misses end in
This scenario is, of course, a long way On the other hand, a look through the data collisions. The industry just about has a grasp
off; autonomy on the roads is in the early on driverless autonomous cars (without a of how safe the average person is by mea-
days, and collective perception is a univer- safety driver on board) reveals that, on average, suring the rate at which people crash, cause
sity research project. However, autonomous a driverless Waymo is involved in collisions injuries and cause fatalities. However, insuffi-
vehicles are already starting to show their once every 52,000 miles. Cruise fairs slightly cient data exists to understand human driving
potential safety benefits. better, with one collision per 63,000 miles. So, safety across the full spectrum of abilities to
they are colliding at roughly twice the rate the same depth as autonomous driving safety.
SAFETY of average San Franciscan drivers. But again, It was important to show with data and
In October 2023, California ordered General this is not the whole story. Out of 33 recorded numbers that AVs could exceed the driv-
Motors to remove its driverless Cruise cars collisions involving a driverless Cruise vehicle, ing standards of the average human driver.
from state roads, stating that GM had “mis- blame can only be attributed to the auton- However, to go beyond this and show that
represented” the safety of the technology. It omous driving system in six cases, which autonomous cars are safer than any human
is clear from this incident and others that AV equates to once per 344,000 miles. Likewise, driver, the benchmark will need to change or
safety claims will always and should always Waymo’s autonomous driving system could evolve. Even beyond near misses, there are
be subject to public scrutiny. This is not to be held responsible roughly for five collisions, many metrics that could be used to better
disparage AVs or their safety, however. Rather, equating to one per 238,000 miles. So by this understand human driving safety, such as how
public oversight should be undertaken not metric, autonomous drivers are once again often drivers go through red lights, how often
with the motivation to eliminate autonomous safer than the average human driver. they tailgate, how often they disobey traffic
cars but with the intent to improve them. Another metric that should be considered rules, how often they exceed speed limits and
Unlike accidents involving human drivers, is the rate at which fatalities and serious inju- many more. With these metrics, the indus-
the intelligence gathered on a single AV ries occur with AVs. Once again, this is lower try could better understand human driving
collision can generate improvements that can for AVs than for human drivers. performance and make solid comparisons to
be rolled out via over-the-air updates to the However, there is a big issue with how build a case that AVs are safer.
entire fleet. In this case, a collision provides this discussion is framed, and it concerns Between sensors, software and safety, it is
an opportunity to prevent that type of colli- the “average” human driver. Clearly, human an exciting time for the automotive industry.
sion from ever happening again, something driving proficiency is a spectrum, encom- The advent of AVs in recent years is an era-
far beyond the capabilities of human drivers. passing those who are incapable of staying defining moment. In 100 years’ time, exam-
By all metrics, the safety of AVs is improv- below the speed limit and off their phones, ples of today’s autonomous cars and robotaxis
ing. The data from California shows as much. the super-vigilant types who have never had will be exhibited in museums as part of the
Going back to 2015, when records were first an accident, and everyone in between. The story of how the automotive industry followed
kept, autonomous cars would barely make it question, then, is with which human drivers a pathway to infallible road safety. ■
1,000 miles without the safety driver behind an autonomous car should be compared. Are
the wheel intervening. Fast forward to 2023, autonomous drivers safe enough when they James Jeffs is senior technology analyst at
and Cruise submitted 576,000 miles worth of are safer than 90% of human drivers? Should IDTechEx.

www.eetimes.eu | JUNE 2024