[log in to unmask]" _mf_state="1" title="null" src="cid:[log in to unmask]" class="" width="44" height="44" border="0"> The SmartDrivingCars eLetter, Pod-Casts, Zoom-Casts and Zoom-inars are made possible in part by support from the Smart Transportation and Technology ETF, symbol MOTO. For more information: www.motoetf.com. Most funding is supplied by Princeton University's Department of Operations Research & Financial Engineering and Princeton Autonomous Vehicle Engineering (PAVE) research laboratory as part of its research dissemination initiative
M. Sena, Oct, 27, "There is no Musings section in this month’s issue of The Dispatcher, and the lead article is long. Fred Dryer, who graciously agreed to review the lead article, The Coming Battle Over Car Electrification, and provided many valuable suggestions for modifications, said he thought it was too long and worried that people would not read it. I said that the 10-15% of those to whom I send the newsletter each month who actually do read each issue would read the article, and the rest might remember that they had received something on the topic if they need it for some other purpose. If you are in the 85-90% group, maybe you could just read The Issue’s Theme: Responsible Transitions on the cover page, as a favor to your editor. I have copied it into the bottom of this mail. I believe you will find it useful for analyzing what is occurring at the COP26 climate meeting starting in Glasgow in a few days." Read more Hmmmm... Another excellent edition and check out the SmartDrivingCars Pod-Cast Episode 239, Zoom-Cast Episode 239 w/Michael Sena, Editor of The Dispatcher, Alain
B.
Templeton,
Oct. 25, "Most
of the biggest
names in
self-driving
cars are
aiming to make
money selling
Robotaxi
service — most
quickly
described as a
self-driving
Uber UBER
-2.2%-style
service where
you can summon
a car with an
app on your
phone and ride
elsewhere for
a reasonable
fee, possibly
combined with
“sharing” in
some form,
such as the
style of
UberPool or
forms of
on-demand
transit.
This is the
plan of Waymo,
Cruise, Amazon
AMZN -3%/Zoox,
Argo AI and
many others.
It was
obviously the
plan of Uber
ATG before it
sold to
Aurora, and
Lyft LYFT
-2.3% L5
before it sold
to Toyota.
It’s even a
big part of
the plan of
Tesla TSLA
+2.3%. While
Tesla intends
to sell
consumer cars,
it promotes
that they will
become part of
the “Tesla
Network” where
customer cars
temporarily
hire out to
give rides,
and also that
Tesla will buy
3-year old
cars as they
come off lease
to put into a
robotaxi
fleet.
Indeed, last
week, Dan
Ammann, CEO of
GM’s Cruise
predicted that
they would see
$50 billion in
sales..." Read
more Hmmmm... Nice article, Brad. I agree. I
really like
his pointing
out that
Chandler may
be great for
test, but the
market is most
challenging
(70% HH have 2
or more cars,
and the mind
set is
completely
focused on the
conventional
car... no one
moves there
that doesn't
love the car
culture.)
I also like
his assessment
of
Uber/Lyft...
How in the
heck aren't
they
profitable
with revenues
of $0.50/mile
and perform
only
internet/SV
type things.
They should be
profitable.
Scalable is a
complete other
question. I
can make a
lemon aide
stand
profitable but
can't possibly
justify a
SiliconValley-type
CapValue.
Their market
penetration is
bounded by the
cost of the
driver. There
simply aren't
enough low
cost gig
workers trying
to make ends
meet. In
their
operation,
scale is
accompanied by
a
monotonically
increasing
driver cost or
(likely...and)
a
monotonically
decreasing
price per
mile. Both
the increase
and dilute the
$0.50 they
currently
take. If
they're not
profitable @
$0.50, how can
they be
profitable at
> $0.50?
At best they
remain a nice
niche
player(combined
they serve way
less than 1%
of the trips.)
or they are
dead.
but I remain
optimistic....
Wrt aTaxis, I
and some
others in NJ
are focused on
New Jersey and
the Trentons
of this
country/world.
We seem to
have a
slightly
different view
of this world
than you in
California.
California...
car culture in
spades! New
Jersey... car
culture, take
it or leave
it! This
difference may
enable an
initial spark
in Trenton to
ignite a
state-wide and
NE region-wide
economically
thriving aTaxi
deployment...
or that is
what I'm
betting on.
Trenton has
70% of HH with
1 or fewer
cars. We have
"rural" poor
in NJ that
spend 80% of
their
disposable
income on a
car, because
they are stuck
living on
inexpensive
land (Where is
inexpensive
land... you
know the
answer...).
Plus aTaxis
are so
infinitely
better than NJ
Transit buses
that NJ's poor
will really
appreciate the
service.
That's what
I'm betting
on. Of course
I then need
the New
Jersians that
look like the
folks in
Chandler and
SV to then
say... "Hey, I
want that
too..." (And I
need to watch
over the
operation to
make sure that
the aTaxi
operator
doesn't
abandon my
Trentonians in
favor of those
SVs).
Again... nice
article. Alain
J.
Bursztynsky,
Oct. 27, "Uber
announced
early
Wednesday that
Hertz, which
just ordered
100,000
Teslas, will
supply half of
that fleet by
2023 to Uber
drivers who
want to rent
the cars...."
Read
more Hmmmm...Either way good for Tesla in that
many more
people can
experience a
Tesla.
Good for Gig workers??? Can they really afford to rent
cars from
Hertz? If
they could,
they would
have in the
past. If they
can't then
either Uber or
Hertz or both
is/are
subsidizing
this
initiative.
That means one
or both are
going to be
more
unprofitable.
Be careful,
there may be
mirrors all
around this
deal. Alain
M. Slovick, Mar. 19, 2021, "Tesla’s Autopilot and Cadillac’s Super Cruise can do some of these things, but both of those systems and others currently on the market require the driver's attention in all situations. With Honda’s level 3 Traffic Jam Pilot, the driver and the system “will share the driving task,"... Read more Hmmmm...What???? Tesla & Cadillac require the driver's attention in all situations. With Honda the driver & the system share the driving task? Interesting use of words largely in the "Elon tradition". Don't say something explicitly that you really can't do (allow the driver to not pay attention). Just use works that leads the reader to imagine that it is OK to do that. .
The automobile industry's business case is founded on
the driver
accepting the
burden of
operating the
vehicle to get
it from point
A to point B
and having
that driver
accept all the
legal and
financial
responsibilities
incurred while
driving from A
to B. One of
the elegant
beauties of
that business
model, from a
driver's
perspective,
both of those
elements,
driving and
liability, are
perceive to
cost nothing
when one is
driving one's
self. If
someone else
is driving,
other than
friends&family,
then these
cost become
explicit even
though they
have, in
reality, not
changed.
Key here is what is not said above... that Honda will
(gladly)
accept the responsibility
for any
bad
eventuality
that derives
from when the
system was
doing the
driving. If
Honda wishes
to accept that
responsibility
then they are
sharing in the
driving and it
is a Level
3. If not,
this is just
like everyone
else's. Not
Level 3. Alain
Date Change: Thursday evening, May 5,
through
Saturday May
7, 2022. Live
in Trenton,
New Jersey.
"Everything" was going well wrt the 5th Summit Nov 18->20; however...
Time is very short, we must focus on the election and the realities of where we remain with Covid really put cold water on some aspects of our vision for November.
Consequently. I've become convinced that it is very much better, that we take our time and reschedule for the first week on May 2022 rather than rush for what isn't as good as could be.
By May we will have received responses to our upcoming RfI for "Equitable ... Mobility in Trenton". We will thus have a better idea on "Who", from the "What & How" communities, "Want & Can" Deploy "Equitable ... Mobility" starting in Trenton with real expectations of scaling throughout New Jersey.
In May the 5th Summit can better achieve its fundamental purpose by allowing all of us in New Jersey to better learn from others around the world the "Whats & Hows" and have the possible "Whos" get a better appreciation of the "Equitable ... Mobility" desires of Trentonians and other New Jersians.
Rescheduling the Summit to be between the RfI and the RfP steps of our Public-Private-Partnership (PPP) Deployment Process, will better enable our Community Engagement initiative to shape our ultimate deployment. We'll thus deliver even better mobility equitably and best improve quality-of-life in Trenton and throughout New Jersey.
Please pencil into your calendar the new dates of May 5 (Thursday evening) though May 7 (Trenton Mobility Festival Saturday). 😎 Please let me know if these dates work for you.
Sorry about the delay, but many will be relieved by this change. Plus, early May is really nice in NJ.😁
Alain
This
5th
Summit is
inspired by
the many
levels of
public-sector,
community and
neighborhood
welcoming and
support that
now exists in
New Jersey for
the deployment of
equitable,
affordable,
high-quality
mobility. This
is now made
possible by
automated
driving
technology
that is
especially
targeted to
serve
those who, for
whatever
reason, don't
have access to
their own
personal
car.
The
Princeton SDC
Summits were
initiated in
2017 to
provide a
venue for the
open
discussion of
how
technology, in
particular
automation,
can be shaped
to improve
mobility of
people and
goods between
and within
cities. Early
on, we
realized that
this form of
mobility could
easily provide
yet another
alternative to
those who are
fortunate
enough to
enjoy one or
many high
quality
mobility
options.
But, more importantly, it became obvious that substantial improvement in quality-of-life and equitable mobility can readily be made available to the un-served and under-served. Those who cannot drive themselves, cannot afford the transport alternatives that exist for them, or who live in areas where, for either economic or other reasons, neither public nor private desirable forms of transport are offered. Furthermore, such initial Operational Design Domains (OODs) can be readily expanded and replicated to allow the vast investments continuing to be made in this technology to actually yield their envisioned societal and financial benefits.
The
recently
completed 4th
Summit set the
groundwork for
these initial
deployments to
benefit under-served communities. Communities with many households
having access
to one or
fewer cars and
with
challenged
transit
alternatives.
We concluded
the 4th Summit
by envisioning a deployment throughout Trenton, NJ, a
community
where 70% of
the households
have access to
one or fewer
cars.
We
believe that
Trenton is a
perfect ODD to
begin to
deliver Equitable,
Affordable,
High-quality
Mobility,
in addition to
being
environmentally
responsible,
safe and
comfortable.
The
opportunity to
expand
throughout
Mercer County
and replicate
this
deployment
scenario
throughout the
State exists.
This
deployment
will serves
as a blueprint
for the future
for many other
"Trentons" of
this world.
The
groundwork set
by the 4th
Summit and the
NJ Autonomous
Vehicle Task
Force has
enabled us to
create a "most
welcoming
environment"
in New Jersey
for creating a
Public-Private
Partnership to
deliver this
enhanced
mobility to
the residents
of Trenton and
all New
Jersians. The 5th Summit will
focus directly
on deployment in Trenton and will take place in Trenton. We also
envision its
expansion
throughout
Mercer County
and its
replication in
and around New
Jersey's other
major
cities.
H. Poser'77, Sept 13, 2020. "Creating Value for Light Density Urban Rail Lines" . See slides, See video Hmmmm... Simply Brilliant. Alain
These
editions re
sponsored by
the SmartETFs
Smart
Transportation
and Technology
ETF, symbol
MOTO. For more
information
head to www.motoetf.com
R. Mitchel, Oct 7, "Compared with so-called advanced driver assistance systems such as Autopilot, a forward collision avoidance system is relatively crude. It is designed to answer one question — is a frontal impact imminent? — and respond to danger by sounding a warning and, if necessary, triggering a subsystem called automatic emergency braking. Unlike Autopilot, which must be selected manually and is available only under some driving conditions, automatic emergency braking runs by default unless manually turned off....
Tesla calls its vehicles “the safest cars in the world,” citing their combination of structural engineering and advanced technology. But when it comes to the forward collision avoidance system, Tesla owners have been reporting problems at a substantially elevated rate compared with similarly equipped cars....
“Teslas are running into stationary objects,” said Alain Kornhauser, who heads the driverless car engineering program at Princeton University. “They shouldn’t be.” If the company’s cars can’t avoid crash scenes marked by flares or traffic cones, he said, “how can you trust anything else they do with Autopilot?”...
One
possibility,
according to
Missy
Cummings, a
former Navy
fighter pilot
who studies
human-machine
interaction at
Duke
University, is
that Autopilot
is designed to
preempt or
suppress
emergency
braking to
minimize
what’s known
as phantom
braking.
“I haven’t
seen the code
to say how
Tesla works,
but I suspect
the AEB is
turned off in
some
situations,”
she said. “If
it were left
on it may
detect what
are called
phantom
objects and
would be
slamming on
the
brakes.”... "
Read
more Hmmmm...
This story is
great and is
not what
anyone else
has written.
Thank you,
Russ, for
doing all of
the research
and hard work
that you put
into this
article.
I agree with Missy, (I haven't
seen the code
either), but,
rest assured,
a perception
algorithm is
part of each
of Tesla's
automated
systems that
"drive" their
cars some of
the time, be
it its forward collision avoidance
system (FCAS),
autoPilot or
FSD. They may
each have its
own or the
best one is
used in all
three, but
each has an
FCAS; else,
Teslas would
never know to
invoke any of
the driving
sub-tasks,
like slow down
or speed up or
stay in the
lane, or don't
hit me, or
... If a
Tesla
perception
system detects
an objects
that doesn't
exist and
locates it as
being in the
lane ahead,
i.e. a
"stationary
phantom object
in the lane
ahead", then
that
perception
system will
signal to the
automated
driving
system... to
slow down and
don't hit it.
If Tesla's
human driver
is paying
attention to
what is going
on and, of
course,
doesn't see
the phantom
object (it is
phantom = not
real), then
the alert
driver is
justified in
losing trust
that FCAS,
autoPilot or
FSD is
reliable and
won't kill.
The erosion of
that trust
leads to
complaints,
demands for
refunds and
substantial
problems for
Tesla.
Tesla
has simply
gotten ahead
of itself in
trying to get
to Driverless
too quickly,
rather than
making sure
that Automated
Emergency
Braking (AEB)
works better
than "good
enough". My
guess is,
Tesla
perception
algorithm
simply ignores
stationary
objects
detected in
the lane
ahead and
those detected
to be to the
side of the
lane ahead.
Ignoring detected stationary objects is perfectly appropriate when following a car ahead. The car ahead didn't crash into that detected stationary object ahead, so the coast is clear! I'll be able to pass under/to the side/over it too! If the car ahead crashes into that object, Its sudden deceleration is readily perceived by the trailing Tesla's AEB. As long as the Tesla has not been tailgating (which a good AEB should disallow), the Tesla should be able to stop in time to avoid crashing into the new pileup ahead. All easy, and likely not the scenario in any of the NHTSA crash investigations.
Not
so easy if the
Tesla is the
lead vehicle,
especially if
the vehicle
that the Tesla
was following
suddenly
changes lanes
and is no
longer
explicitly
confirming
that the
Tesla's road
ahead is
traversable,
It is now the
Tesla's job to
determine if
it can pass
under a
stationary
object in the
lane ahead.
That is simply
not easy
to do
reliably. Not
easy to
determine the
clearance
under an
overpass/sign/traffic
light/tree
canopy while
approaching
said
overpass/...
at any
significant
speed. If the
object is
classified as
an overpass/sign/traffic
light/tree
canopy,
the chance are
really good
that "passing
under" is a
breeze.
However, if
classification
of the object
is uncertain,
then all bets
are off.
I strongly suspect that Tesla's perception algorithm disregards all stationary objects ahead when leading as well as when following. NHTSA has to tell Tesla to not do that any more!!! Tesla must go back, essentially to the beginning, and figure out how to reliably determine if it can pass under, beside or over stationary objects detected in the road ahead. Alain
A.
Hawkins, Sep
30, "Waymo and
Cruise, two of
the leading
autonomous
vehicle
companies in
the US,
received
permits from
the California
Department of
Motor Vehicles
to offer rides
to passengers
in their
robotaxis.
But while
Cruise was
approved to
give rides in
its fully
driverless
vehicles
without safety
drivers, Waymo
only is
allowed to
deploy its
autonomous
vehicles with
a human
monitor behind
the wheel. In
order to give
rides to
paying
passengers in
its fully
driverless
vehicles, as
it does in
Arizona, the
Google spinoff
would need to
apply for an additional
permit from
the California
Public
Utilities
Commission.
... "
Read more Hmmmm...
Congratulations Kyle, Robert and everyone else!!!! This is a non-trivial
accomplishment!
Given all of the additional knocks
on your door
that will
naturally come
your way, we
hope that
you'll keep us
in mind. We
here in NJ
have assembled
an enormously
welcoming and
realistic
environment
for Deployment
to a customer
base that will
fundamentally
benefit and
cherish the
Equitable,
Affordable,
High-Quality,
Safe Mobility
that is
delivered by
your
Driverless
Technology.
Hopefully
you'll divert
a little
bandwidth to
our upcoming "5th Annual
SDC Summit",
New Jersey's
next step in
our process to
help you and
possibly
others get to
where you've
gotten in
California.
We are new
kids on the
block, but
we've really
gotten our act
together to
work with you
and others to
catch up
quickly and
really improve
the
quality-of-life
for many here
in New
Jersey, and
the rest of
the NorthEast.
Again... Congratulations! So pleased and so well deserved! Alain
229 Sherrerd Hall Princeton University Princeton, NJ [log in to unmask] 609-980-1427 (c) |
[log in to unmask]" class="" width="79" height="79" border="0"> |