zoox
concept for an AV (image courtesy of zoox)
February 2015
From the Editors
Competition seems to be a key theme emerging
this month. The potential of Autonomous Vehicles (AVs) is
more widely recognized and the competition to be a leader in the
technology and the various sectors and niches is starting to heat
up.
Competition was very evident at CES, with Mercedes, Audi, BMW,
Nissan and others competing for bragging rights; more below.
However, the most obvious competitive clash in the AV space in the
last few weeks is that between Uber and Google. In the past,
we have covered the large investment by Google Ventures into Uber,
the fascination that Travis Kalanick (CEO of Uber) has with AVs,
their vision to "make car ownership go away", and the
joint acknowledgement that Google and Uber have discussed the
subject. Many of us have speculated on how powerful a
mobility service partnership between Google and Uber would be.
This seemingly happy accord appears to have been shattered by a
story
posted by TechCrunch on February 2, 2015 stating that Uber has
hired most of the Carnegie Mellon University (CMU) Robotics
Institute staff in order to kick-start autonomous taxi fleet
development. CMU has given us the impression that they were
the leading academic-based AV developer, but they have now been
brought into the fold of the private sector.
Both Uber and CMU
released official statements confirming their "strategic
partnership", however a comment on LinkedIn suggests that
this was more likely a well played business deal by Uber.
By coincidence (?), a related
story by Bloomberg on the very same day stated that the Uber
board was shown screenshots of a Google ride-sharing app currently
being used by Google employees.
What we do know is that some very deep-pocketed companies are
investing heavily in an AV future where private vehicle ownership
will be replaced by a service model. Given the overall
socio-economic benefits that this can bring, we believe that
competition in the AV space will be a good thing in the long term.
The
Automakers (and Tier1s and future OEMs)
Daimler/Mercedes
Benz at CES the F 015 was unveiled to great fanfare. A highly
innovative and futuristic concept, the F 015 promoted some very
interesting discussion. Our Paul Godsmark felt it covered the
full spectrum of amazed from "great!" to
"why?".
Audi decided to let the car drive itself to CES in Vegas,
thus generating
the headline "I Rode 500 Miles in a Self-Driving Car and
Saw the Future. Its Delightfully Dull".
Nissan announced
that they have teamed up with NASA to develop AVs. Given
that Elon Musk famously wants to die
on Mars (just not on impact), we wonder if he will be being
driven by a Tesla or a Nissan when hes there…
Nissan also demonstrated
their autonomous electric Leaf to a journalist by having it
drive around Sunnyvale. The Nissan system uses sparse maps
from third parties, in contrast to Googles approach of using very
detailed maps.
BMW at CES demonstrated their self-parking i3s, but then
dropped the bombshell that they estimate this feature wont
be available for another 7 years. Are they aware of what
their competitors are saying?
GM seem open to having a discussion
with Google on the subject of AVs.
North American International Auto Show (Detroit) it was
very clear that AVs
have moved to the forefront.
Bosch outlined their 10-year
path to a door-to-door AV.
Google "Google's
Self-Driving Car Pals Revealed" which includes
"Continental, Roush, Bosch, ZFLS, RCO, FRIMO, Prefix and LG"
who helped build their prototype. For those not familiar
with Googles technology this recent article "Roadtesting
Googles new driverless car" is a great introduction.
Ford "Ford
CEO Expects Fully Autonomous Cars In 5 Years", but
it won't be Ford that makes them in …that time frame
"because our approach is when we do, we want to make sure
that it is accessible for everyone and not just let's say luxury
[car] customers."
Transit
/ Transportation
An analysis of "Taxi
vs Public Transit in major US cities" suggests that in
most cities, it is already quicker to catch a taxi. With the
advent of shared automated taxi fleets, the costs of taxis will be
significantly reduced, resulting in a greater modal shift from
transit than is already occurring.
AV
Trials
Singapore "Testing
of driverless vehicles at one-north expected to start in March".
The trials will be conducted on a 6km network of roads that
encompasses both light and heavy traffic routes.
Socio-Economic
Impacts of AVs
"I
Find Your Lack Of Faith In Autonomous Cars Disturbing" a
very enthusiastic article which conveys many of the huge key
issues which AVs pose.
Government
/ Regulatory
The Ministry of Transportation of Ontario
(MTO) continues to have an interest in allowing testing of AVs on
Ontario's public roads. The first stakeholder consultation
meeting was held in February 2014. The process was
interrupted by the Ontario election in mid-2014. Then, in
late 2014, a number of focused consultation meetings were held
with a range of stakeholders to address more specific issues.
CAVCOE's Barrie Kirk contributed to both rounds of the
consultation process. There is no word from MTO on when their work
will be completed, but we can expect their continued effort on
this front as highlighted in the Minister
of Transportation's publicly available 2014 Mandate letter.
US DOT: "Beyond
Traffic: US DOT's 30 Year Framework for the Future" was
launched at Google's headquarters in Silicon Valley. We are
very pleased that Secretary Foxx has recognized that Google may
play a very important role in this conversation. We are
however concerned with the nature of the conversation. It
underscores critical decision points facing the country, by means
of data driven analysis, research, expert opinions and public
engagement; means that it will take time and is not intended to
arrive at prescriptive solutions. Meanwhile the date at
which AVs will be deployed in earnest is getting closer by the
month and USDOT still doesnt allow for the impacts of AVs in any
of their expensive infrastructure proposals. The Draft
Report "Beyond
Traffic: Trends and Choices 2045" is very interesting,
but to find references to AVs you need to search for
"automation", "automated",
"autonomous", "self-driving" and
"driverless".
California DMV held a public workshop to discuss proposed
regulations and the technical safety standards for AVs. The press
headlines stated that Google were warning CA DMV not to
over-regulate AVs and that they, Google,
could do a much better job on their own. However a more
balanced view emerges from an eye-witness account sent to Prof
Alain Kornhauser who included in one of his regular emails:
I was at the CA DMV hearing and was sitting right next to our
friends at Google. The reporting around this is, as usual, was
taken a bit out of context. Brian Soublet from the DMV was
very clear about the fact that the DMV lacked technical expertise
to properly evaluate self-certification by OEM's, and the Google
rep was simply repeating that as part of his argument."
The elephant in the room was that the DMV was tasked by the state
with the incredibly difficult task of deciding what products to
allow on public roads despite the fact that they haven't the
budget, expertise, or resources to do so. In fact, we
weren't able to identify anyone who was really capable of doing
this, which is why self-certification seemed like the only logical
choice.
"Google
Says Snow More of Self-Driving Car Snag Than NHTSA" was
the headline around comments made by Chris Urmson of Google when
he pointed out that NHTSA could ban AVs, but Google didnt think
that NHTSA needs to grant permission ahead of time. Google also
reiterated its desire for partners to help realize its goals.
(Note the comments from GM above).
North Dakota "Lawmakers opt for study,
not regulation of unmanned vehicles". The CNATCA
autonomous-friendly corridor
concept was recommended for further study.
The Dutch government has approved the large-scale
testing of AVs. Dr. Alexander Hars asks: "Netherlands
first to operate a self-driving shuttle in public traffic?"
to which the answer might be "no" depending on how you
view CityMobil2s
trial in La Rochelle, France, the Singapore
AV Initiative at One-North, or a possible Google trial (in
Mountain View?).
Germany developments are summed up by this headline "Germany
Declares "Digital Sovereignty", Snubs Google's
Self-Driving Cars
UK - The government wants the UK
to become a world leader in driverless technology. It
will publish a code of practice in the spring which will allow the
testing of autonomous cars to go ahead. The
official UK government press release is just as bullish.
"UNECE
(United Nations Economic Commission for Europe) to discuss first
international regulatory steps concerning automated-driving."
South
Australia wants to become the first state (in Australia)
to pass rules that would allow driverless cars on its roads,
possibly within years.
Research
The
University of Michigan believe that they have developed a lower-cost
navigation system for AVs using localization based on 3D
interpretation from cameras as opposed to the more expensive LiDAR.
The University of Michigan has also completed its "M
City" which includes a network of roads and
infrastructure for testing
autonomous and connected vehicles.
The "European
Roadmap: Smart Systems for Automated Driving" clearly
lays out a road-map and a time-line for AVs. But we note
that they dont envisage highly automated driving to be deployed in
cities until around 2030. We remind readers of Googles
consistent re-statement of their intentions to have their
self-driving technology in public hands by 2017-2020.
A study by Imperial College, London indicates "How
Driverless Cars Could Make Traffic Dramatically Worse".
Apparently if AVs accelerate and decelerate smoothly and
comfortably at rates similar to trains, then congestion is
dramatically increased. Redditor "secondlamp" summed
it up thus: "If we make SDCs slow, SDCs will be
slow."
"Characterization
of the OPAL Obscurant Penetrating LiDAR in various Degraded
Visual Environments" is interesting as this high-power LiDAR
can still sense objects in snow, dust and fog. It remains to
be seen if it is even possible for much lower-power units to be
developed with similar capabilities.
"THz
sensing in front of Vehicle (Ground Profiling)" is of
potential value to AVs.
"A
single self-driving car could take care of multiple people's daily
driving needs, so households might not need as many
cars", University of Michigan researchers say.
But the way this study is framed, we might also expect more cars
on the road during peak periods as it assumes household vehicle
ownership and empty vehicle trips between serving family members.
We would like to see a similar study that considers the
possibility of shared AV fleets similar to what Uber are expecting
to provide.
Other
AV Articles
The
Boston
Consulting Group estimates that AV features could represent a
$42 Billion market by 2025. But their slide
deck clearly shows their assumptions are based on private
vehicle sales and that high hardware costs will limit market
penetration to about 25%... We suggest that they conduct a further
analysis for a shared AV (SAV) scenario where large fleets of SAVs
replace private ownership, as Uber and Google are proposing. The
automotive analysts dont seem to be factoring in the
socio-economic impacts that occur when the vehicle can move people
and things. See the report
co-authored by CAVCOE, the Conference Board of Canada and the Van
Horne Institute for more background to this.
"Expect
driverless insurance tussle: Allstate CEO" is a headline
that further indicates how the AV discussion has moved into the
boardrooms of many companies that are anticipating direct impacts.
Allianz
are also talking about AVs and insurance.
Steve Jurvetson (one of Silicon Valleys most prescient Venture
Capitalists) was all
smiles when visiting
zoox. Interesting…
Brad Templeton has been pondering "Will
robocars use V2V at all?" and suggests that "the
answer is not certain." Food for thought.
"Uber CEO explains his companys highly ambitious goal to end
car ownership in the world".
Did anyone think of sending an AV
themed Valentines card?
Technology
At CES NVIDIA unveiled
their new Drive PX which has been specifically
developed for AVs. It combines supercomputer-like
capability and deep neural learning.
With reference to the NVIDIA unveiling above, Mobileye Co-Founder,
Chairman & CTO Prof. Amnon Shashua talked about the future of computer
vision and automated driving at the 2015 Deutsche Bank Global
Auto Industry Conference and put a strong case for the potential
of the Mobileye tech.
Upcoming
AV-related Events
April 23-24, 2015: Insurance
Telematics Canada 2015; Toronto, Ont.
May 4-7, 2015: AUVSI's Unmanned
Systems Conference and Trade Show; Atlanta, Georgia
May 5-6, 2015: 4th
Annual Wavefront Summit; Ottawa, Ont.
May 24-28, 2015:
ITS Canada's 2015 Annual Conference and General Meeting;
Gatineau, QC.
June 16-19, 2015: CCMTA's Annual
Meeting; Whitehorse, Yukon.
June 17-18, 2015: Autonomous
Vehicle Test & Development Symposium 2015; Stuttgart,
Germany.
July 21-23, 2015: AUVSI / TRB Automated
Vehicles Symposium 2015, University of Michigan, Ann Arbor MI,
USA.
October 5-9, 2015, ITS
World Congress, Bordeaux, France
|