Friday, August 23, 2019

https://www.princetondiary.com/smartdrivingcar/7.35-Myths-82319

  Five myths about autonomous vehicles

J. Browne, Aug 16, "utonomous vehicles are the future. Self-driving cars could change our lives, heralding an era of greater convenience, improved productivity and safer roads…."  Read more   Hmmmm….  Actually much of this opening sentence is a myth… It doesn’t take Self-driving or Driverless to have automation technology yield safer roads.  It takes safe-driving technology that works, like Automated Emergency Braking (front and rear)…  And … are we really going to do our "manufacturing or service job " (increase "productivity") if we don’t have to do the work of driving anymore???  Of the few "riding shotgun to work" what percentage are doing work while riding shotgun?  Certainly less than 10%.  Less than 1%?  So much for productivity improvements 

If we get to Driverless, then the myths aren’t myths.  There will be fewer private cars, downtown congestion will be reduced, the environment will be saved, the insurance industry’s gross revenues will go down substantially  (but their profits will go up) and AVs are already safer than humans that text and/or are "under the influence" while driving. 

If we don’t get to Driverless, then we’ll remain with "Do-it-yourself private mobility" that will include Self-driving assistance.  Armed with that form of personal mobility, then all the myths are myths: More private cars …   and the policy implications are clear.  See: J. M. Greenwald, A. L. Kornhauser "It’s up to us: Policies to improve climate outcomes from automated vehicles".   Also, to have a proper perspective of the role of transportation and car/"FordF150s" in greenhouse gas emissions see…  M. Sivak, Aug 22, "Increased relative contribution of medium and heavy trucks to U.S. greenhouse gas emissions"   Alain

   Smart Driving Cars Podcast Episode 121 -Ken Pyle

F. Fishkin, Aug 22 , "Daimler and Bosch hold a community meeting in San Jose as they ready plans for autonomous vehicle testing. Community board member Ken Pyle joins Princeton’s Alain Kornhauser and co-host Fred Fishkin. Plus…Waymo, Tesla and more."  Just say "Alexa, play the Smart Driving Cars podcast!".  Ditto with Siri, and GooglePlay …  Alain

  Waymo is making some of its self-driving car data available for free to researchers

A. Hawkins, Aug 21, "The data collected by self-driving cars used to be a closely guarded secret. But recently, many companies developing autonomous driving systems have begun to release their data to the research community in dribs and drabs. The latest to do so is Waymo, the self-driving unit of Alphabet, which today is making some of the high-resolution sensor data gathered by its fleet of autonomous vehicles available to researchers.

Waymo says its dataset contains 1,000 driving segments, with each segment capturing 20 seconds of continuous driving. Those 20-second clips correspond to 200,000 frames at 10 Hz per sensor, which will allow researchers to develop their own models to track and predict the behavior of everyone using the road, from drivers to pedestrians to cyclists…."  Read more   Hmmmm…. Fantastic.  Data set is available @: https://waymo.com/open/  Alain

  Meet Tesla’s self-driving car computer and its two AI brains

S. Shankland, Aug 20, "Designing your own chips is hard. But Tesla, one of the most aggressive developers of autonomous vehicle technology, thinks it’s worth it. The company revealed details Tuesday about how it fine-tuned its AI chip design so two of them are smart enough to power its cars’ upcoming "full self-driving" abilities.
Tesla Chief Executive Elon Musk and his colleagues revealed the company’s third-generation computing hardware in April. But at the Hot Chips conference today, chip designers showed how heavy optimizations in Tesla’s custom AI chips dramatically boosted performance — a factor of 21 compared to the earlier Nvidia chips. As an added bonus, they’re only 80% the cost, too….

Each Tesla computer has two AI chips, a redundant design for better safety, Venkataramanan said. There’s also redundancy in the chips’ power supplies and data input feeds, too. Even the car’s cameras are on two separate power supplies to guard against failures….

Each chip makes its own assessment of what the car should do next. The computer compares the two assessments, and if the chips agree, the car takes the action. If the chips disagree, the car just throws away that frame of video data and tries again, Venkataramanan said. That’s one of the reasons Tesla wanted powerful AI chips that could handle such a high frame rate for video. …"  Read more   Hmmmm….  Nice that they have two systems running in parallel.  Since they are the same systems, one would expect that they would give the same result; however, they could each be making the same mistake. Woops! 

 

Also, when they don’t agree, hopefully the input and output of each is logged and some human tried to figure out why. 

 

Also, it is a shame that it processes frames one at a time, starting over each time.   As opposed to using what it assessed/"learned" from the previous frame(s)  to better and more quickly assess what is going on in the newest frame.  Driving safely down a road shouldn’t/doesn’t involve starting each cognitive cycle clueless about what was just experienced.  Alain

  A Historian’s Take on the Future of Cars

N. Sciple, Aug 18, "In today’s special episode of Industry Focus: Energy , host Nick Sciple chats with historian Dan Albert about the subject of his newest book, Are We There Yet? The American Automobile Past, Present, and Driverless . In the hourlong interview, Dan illustrates what America looked like before cars, how cars changed the landscape and society, and what history can tell us about the future of driving. Tune in and learn why electric vehicles are a lot older than most people realize, and why the diesel engine took over for so long; how infrastructure did and did not play into the car’s mass adoption, and what that means for the future; what drivers will lose if self-driving overtakes traditional driving; how personalization of cars has evolved over the decades; where companies like Uber (NYSE: UBER) and Lyft (NASDAQ: LYFT) fit into the American automotive tapestry; and so much more…. "  Read more   Hmmmm…. Listen to PodCast.  Alain 

  Lyft’s IPO is outperforming Uber’s in one crucial way

A. Griswold, Aug 19, "It’s a big day for Lyft employees. When trading begins on the Nasdaq this morning, they can finally sell their shares.

Like most companies that go public, Lyft included a lock-up period in its IPO terms that barred company insiders—employees, management, and pre-IPO investors—from selling their holdings for 180 days from the day of the offering. The lock-up was supposed to last until Sept. 24, but the company said in its recent quarterly report that it had pushed that date to Aug. 19 because the original date conflicted with a routine “blackout” period that limits trading before the end of the quarter.

With the end of the lock-up, Lyft expects nearly 258 million shares of its Class A common stock to become eligible for trading, out of a total of around 280 million, or 341 million on a fully diluted basis that includes shares that have yet to be converted. Lyft co-founders Logan Green and John Zimmer and chief financial officer Brian Roberts, who collectively own about 5.6% of the company’s stock, have told equity analysts they don’t plan to sell when the lock-up ends. But that still leaves a lot of shares to flood the market, putting downward pressure on Lyft’s stock price…."  Read more   Hmmmm….Very interesting. Uber short volume, Lyft short volume , Tesla short Interest  Alain

  To Beat Uber And Lyft, This Startup Vows To Give Its Drivers The Full Fare

K. Cai, Aug 22, "Uber and Lyft have handily deflected any competition so far. Together, the two ride-hailing giants capture more than 98% of market spending, according to analytics company Second Measure.

Commissions for each ride comprise a significant revenue stream for both companies to the long-standing anguish of many drivers. Uber says it takes 25% of every ride, and while Lyft is more coy about its cut, the rates are estimated to be comparable. However, factors such as ride distance can cause the commission to balloon above 40%—a major reason why drivers for both companies organized a strike in May. When asked in 2015 why Uber was raising its commission rates, a then-executive reportedly said: “Because we can.”

One startup with backing from Silicon Valley accelerator Y Combinator wants to challenge the duopoly by eliminating the commission. Nomad Rides says its platform charges riders 20% less while drivers earn 20% more, paying Nomad a flat monthly subscription—currently, $25 per month. The startup launched its app at Indiana University Bloomington in February…."  Read more   Hmmmm…. Is there no good news anywhere for Lyft/Uber?  Alain

  What Will It Take for Uber to Become Profitable?

D. Wessel, Aug 20, "Uber has to fix its business model to make it profitable in a sustainable way,…"  Read more   Hmmmm…. The problem for Uber/Lyft is much more fundamental.  Mobility-as-a-Service is fundamentally labor intensive… whether it be done with a chauffeur, taxi driver or gig worker.  Most of the revenue goes to labor.  Unfortunately the demand for mobility only affords very few trips that can support a living wage.  That’s why most of personal mobility is "Do it yourself".. we walk, we bike, we drive our own cars.  When we aren’t willing to "Do it ourselves" then we have to share the driver, big time… get into a packed train, bus or airplane.  So the only opportunity for Uber/Lyft to achieve "network effects" is to cut out labor.. the driver… autonomousTaxis.  

 

At the size that they are now, neither Lyft or Uber are close to break-even.  If they grow, it gets worse (unit labor charges go up and average revenue goes down, fundamental supply – demand).  Driverless in a sufficiently large Operational Design Domain to substantially increase their size in not going to happen soon enough to save them as they exist today.  They were just too early.

 

Their only "survival" option is to downsize by a factor of at least 10, reduce their valuation/stock price by a factor of at least 10.  Become a "nice business" for 5-10 years and wait for Driverless aTaxis to become a reality and start this all over again. 

 

Driverless is a necessary condition to make this into a network non-labor intensive business.  Alain

New York City’s school buses will now be automatically routed and tracked using Via’s algorithm

E. Anzilotti, Aug 21, "The yellow school bus is an iconic feature of the American streetscape—their rumbling presence through neighborhoods signals the end of summer and the beginning of a new school year for kids. But as nostalgia-inducing as they are, yellow school buses are not exempt from the wave of disruption sweeping transit in the U.S.

The ride-hailing company Via is licensing its technology to the New York City Department of Education in an attempt to modernize the city’s entire school bus system. Through the Via partnership, school buses will be able to more quickly adapt routes to respond to kids’ needs, and the tech platform will provide GPS tracking and real-time updates to parents and students as they navigate school transit, says Via cofounder and CEO Daniel Ramot.

Instead of offering mainly personal rides like Uber and Lyft, Via is built on a pooling model that it says sits between private car trips and public transit. When a passenger requests a ride through the Via app, the algorithm matches it with other passengers in the area that are all traveling to the same general location (or at least in the same direction). It then designs the most efficient route for a single vehicle to pick them all up and deliver them….."  Read more   Hmmmm….  Very nice.  Alain

  Deadliest Summertime Highways In The U.S. Will Be Recast Via Self-Driving Cars

L. Elliot, Aug 18, "… Those are considered the top five deadliest summertime highways, according to a recent analysis of National Highway Traffic Safety Administration (NHTSA) data on highway fatalities in the United States… Why are those particular highways so deadly?…  Getting human drivers to become better at driving is not an easy assignment. Humans tend to resist change. They tend to get bored or distracted while driving. They get upset and allow their emotions to carryover into their driving decisions. And so on….

I’m not going to tackle herein the matter of how semi-autonomous cars are going to impact our deadliest highways situation and instead focus on the true self-driving driverless cars (which we don’t yet have, and for which we barely have Level 4’s, a constrained version of full autonomy, doing limited public roadway tryouts)…."  Read more   Hmmmm…. Interesting read.  Alain 

  Uber Whistleblower: Autonomous Vehicles Need New Safety Metrics, Aren’t Really Any Safer

M. Posky, Aug 19, "Over the past year the automotive industry has carefully walked back the expectations surrounding autonomous cars. Yet pretty much any change in rhetoric constitutes retracted goals. With numerous companies predicting self-driving fleets of commercial vehicles before 2021, the bar couldn’t have been set much higher.

A lack of progress is partly to blame. However, a bundle of high-profile accidents have also shaken public trust — especially after it was found that Uber whistleblower Robbie Miller was trying to alert the company to issues with its self-driving program just days before one of the company’s autonomous Volvos was involved in a fatal accident with a pedestrian.

That’s not the half of it. In April, Miller released a study claiming self-driving vehicles were actually recording incident rates higher than that of your typical motorist. …  In his study, he alleged that too many automakers and tech firms rely on misleading benchmarks, reframing setbacks, and focus too much on the amount of testing vs technological progress.

“The miles per disengagement metric is a bad metric for measuring progress and is not meaningful in terms of safety.… I sort of disagree…  It is a good metric within a characterized Operational Design Domain (ODD).  It is a bad metric when independent of the ODD.  The issue is where and under what circumstances the disengagement occurs.  For some location/circumstances that are commonly encountered in providing mobility, for example neighborhood streets with numerous intersections, disengagement rates need to be very small because most trips either begin or end in residential neighborhoods.  Similarly for commercial neighborhoods. 

Low disengagement rates on well designed, limited access highways under light traffic conditions in good weather are by themselves not relevant unless they are combined with low disengagement rates in residential and commercial neighborhoods.  High disengagement rates in bad weather or certain locations serve to preclude those locations/circumstances to be included in the ODD, and thus precluding the delivery of mobility services requiring the use of those locations in those circumstances. 

The competition is not to have the lowest Disengagement rate, but instead to have a low enough Disengagement rate for an ODD that enables mobility services to be offered to the set of trips encompassed by that ODD .  So in the end, Disengagement rates are only relevant within the context of what you are trying to accomplish: …  to whom, when, where are you trying to provide mobility.  We are nowhere near, and may never be, doing it for everyone, everywhere, every time. 

The big question is: What is the ODD that has demonstrated a sufficiently low Disengagement rate AND delivers sufficient improvement in quality-of-life to enough trip makers to justify the risks to those emerged in that ODD associated with that non-zero Disengagement rate?"     

He does, however, take time to praise his current employer in his industrial analysis — leaving us wondering how much of it was a sales pitch for Pronto vs a call for sanity. "  Read more   Hmmmm…. None of this is simple.  See comments above.  Alain

  Starship Technologies raises $40M, crosses 100K deliveries and plans to expand to 100 new universities

D. Etherington, Aug 20, "Starship Technologies invented the category of rolling autonomous sidewalk delivery robots, and to date, the company has made over 100,000 commercial deliveries on behalf of customers. The milestone comes as Starship adds $40 million in Series A funding, bringing its total funding to $85 million. When it announced an additional $25 million in June 2018, Starship was also piloting its first university deployment – and now the company has a plan to expand to 100 university campuses over the next two years based on the strength of that pilot…

This vertical focus on post-secondary schools will see Starship robots deployed at the University of Pittsburgh today, and Purdue University in Indiana on September 9, with many more to follow. Starship’s ambitious goal is to deploy at 100 schools within the next two years, as mentioned, and it’s going to be using this funding in pursuit of that expansion. "…"  Read more   Hmmmm… Progress!  Alain

   DoorDash acquires autonomous driving startup Scotty Labs

M. Dickey, Aug 20, "DoorDash  has been on an acquisition tear of late, with Scotty Labs as its latest target. Terms of the deal were not disclosed, but this comes after DoorDash acquired Caviar in a deal worth $410 million.

Scotty Labs,  a tele-operations company that is working on technology to enable people to remotely control self-driving cars, raised a $6 million seed round from Gradient Ventures, with participation from Horizon Ventures and Hemi Ventures, last March. The startup had previously worked with Voyage for its self-driving cars in retirement communities.

“Our core belief at Scotty has always been that Autonomy + Remote Assistance is the future,”… "  Read more   Hmmmm….  Yes, but where is the "autonomy" piece?  "remotely control self-driving"  is an oxymoron.  Alain

  The People of Central Park West Want Their Parking Spaces (Sorry, Cyclists)

J. Barron, Aug 18, "The roadwork notices were tied to trees and signposts on a Friday along one of the world’s best-known streets. The following Monday, crews arrived on Central Park West and began setting up a protected bike lane, which eliminated about 200 parking spaces in an almost mile-long stretch.

The next day, representatives for a luxury Central Park West condo went to court to block the bike lane, opening another front in the increasingly pitched battle between bicyclists and motorists for precious space in the nation’s largest city. …"  Read more   Hmmmm… Can you imagine what will happen once Driverless cars appear unannounced on neighborhood street?  Why are neighbors the last to be engaged when change is contemplated?  Listen to Smart Driving Cars Podcast Episode 121 -Ken Pyle.  Alain

  Rain Testing in the Sunshine State

Waymo team, Aug 20, "Both human-driven and self-driving cars need to drive in many different weather conditions. Waymo is designing and testing our system to work safely and reliably in the toughest environments. Over the last few years our testing has taken us to snowy Novi, Michigan, rainy Kirkland, Washington, foggy San Francisco, and of course those dusty haboobs in Phoenix, Arizona. This month, Florida joins the list: we’re bringing Waymo vehicles — including both our Chrysler Pacificas and a Jaguar I-Pace — to the state to begin heavy rain testing…

First, we’re spending several weeks driving on a closed course in Naples where we will rigorously test our sensor suite — which includes lidar, cameras, and radar — during the rainiest season in the south. Later in the month, we’ll bring our vehicles to public roads in Miami. They’ll be manually operated by our trained test drivers which will give us the opportunity to collect data of real-world driving situations in heavy rain. Additionally, Florida residents will start seeing a few of our vehicles on highways between Orlando, Tampa, Fort Myers and Miami as we learn about Florida roads…."  Read more   Hmmmm…. Interesting.  Maybe in the winter, they’ll bring a few to New Jersey.  We have rain and a little bit of snow as well as a real need for improved shared mobility for many.  Alain

  70 Billion Hours Of Driver Time To Be Freed Up By Self-Driving Cars

L. Elliot, Aug 18, "Americans spend about one hour a day driving and travel approximately 32 miles per day doing so, according to recent analyses. Collectively, this amounts to nearly 70 billion hours of driving time annually in the United States alone.  That is 70 billion hours of humans gripping the steering wheel, keeping their eyes on the road (hopefully!) and otherwise undertaking the heady and life-or-death task of driving a car.

Here’s a question to consider: what will happen to those billions upon billions of driving hours once we have truly self-driving driverless cars available?…."  Read more   Hmmmm…. Certainly a good question. I don’t buy that "privacy" is a fundamental desire.  We are social animals that can readily create privacy even if others are around (certainly what many do with smartPhones and head sets) and I don’t really care if someone sees me picking my nose.  Alain 

Tesla postpones Full Self-Driving price increase until V10 and Smart Summon

F. Lambert, Aug 20,  "Tesla was planning to increase the price of its “Full Self-Driving” package last week that was supposed to coincide with an update to its Autopilot features, like the new Smart Summon, but it now won’t happen until September/October.
Earlier this year, Tesla killed the Enhanced Autopilot package and unbundled some features to create a base Autopilot package, which they made standard on all Tesla vehicles.

They added the features to the Full Self-Driving option and increased the price to $6,000 if ordered before delivery — $8,000 if ordered after delivery…"  Read more   Hmmmm… "Smart Summon" is a BAD/stupid idea at this time, even if its use is restricted electronically to private property.  We all know that it wont be long until some "13 year old" hacks it to summon a Tesla "down my street".  When that happens, things are going to get very UGLY.   Alain

  First on-demand autonomous shuttle trial opens to the public

IT, Aug 20, "Singapore’s first on-demand autonomous shuttle public trial has been officially launched at Sentosa, marking a major milestone since on-road testing of the autonomous shuttles began in June 2018. The public trial will run from 26 August to 15 November 2019. Members of the public visiting Sentosa will be able to hail an autonomous shuttle via the Ride Now Sentosa mobile app or at kiosks along a 5.7km route to bring them to selected destinations on the island, such as Siloso Point, Beach Station, Palawan Beach, Tanjong Beach and Sentosa Golf Club…." Read more   Hmmmm… Unless they get to a point that they remove the attendant, this is just another shuttle bus with expensive gizmos attached.  Alain

Half-baked stuff that probably doesn’t deserve your time 

  China’s Ninebot unveils scooters that drive themselves to charging stations

Y. Yang, Aug 16, "Segway-Ninebot Group, a Beijing-based electric scooter maker, on Friday unveiled a scooter that can return itself to charging stations without a driver, a potential boon for the burgeoning scooter-sharing industry…. controlled remotely from the cloud, …"  Read more   Hmmmm…  Segways with SegHoles on them was bad enough on our sidewalks, now an invasion of Segways operated by remote SegHoles sauntering down our sidewalks in search of energy.  Can’t wait for that.  Alain

China dedicates a mountain highway to test self-driving cars

J. Zhang, Aug 19, "China has dedicated a mountainous stretch of highway in eastern Shandong province for testing connected, self-driving vehicles as part of the country’s efforts to challenge the US in autonomous driving technology.
The local Qilu Transportation Development Group earmarked a 26-kilometre-long (16 miles) highway for autonomous driving and vehicle-to-infrastructure communication, in cooperation with partners including Chinese telecom gear provider Huawei Technologies and China Mobile, one of the country’s largest telecommunications network operators, according to the official Xinhua News Agency…."  Read more   Hmmmm…. OK.  This is all about testing communications equipment and not so much about useful testing of automated driving since "proving that automated driving is safe on this mountain road" only extends automated driving’s Operational Design Domain (ODD) to remote mountain roads which does little to extend the market of trips that could be served by such vehicles.     Alain

  Advancement in the Auto Industry Depends on Infrastructure Spending

B. Schimmel, Aug 18, "Automakers are increasingly investing time and research into creating driverless vehicles, but a Kentucky expert says the new technology could face hurdles if the necessary infrastructure isn’t created.

President Trump and members of Congress have repeatedly tried and failed to make a deal on an infrastructure spending package. According to a report card from the American Society of Civil Engineers, Kentucky’s roads are improving, but there’s still about six billion dollars’ worth of unfunded projects…." Read more   Hmmmm….I disagree. Alain


 C’mon Man!  (These folks didn’t get/read the memo)

Researchers look into keeping autonomous vehicles from becoming mobile vomitoriums

D. Coldewey, Aug 20, "If you’re like me, and I’m just going to assume most of you are, motion sickness is a serious consideration on any car trip where you’re not driving. So what are we supposed to do in self-driving vehicles? Researchers are finally looking into this question with an experiment designed to see just what makes people like us so sick…."  Read more   Hmmmm…. Oh my goodness, someone might be riding in a car rather than driving it.  There’s been nothing but single occupant cars in Michigan.  What must it be like to actually just ride in a car???  I never realized that ride-sharing caused my ’55 Chevy to become a vomitorium.   What????  C’mon Michigan!!!  Alain

  Amazon Forecast hits general availability

K. Wiggers, Aug 21, "Image Credit: AWS
The ability to forecast events at scale, given a set of variables, is something most companies would find useful. So Amazon is aiming to make prediction more accessible with a fully managed service called Forecast that uses AI and machine learning to deliver highly accurate forecasts. It’s now generally available through Amazon Web Services …

As Amazon explained in a press release, Forecast — which is based on the same technology the Seattle company uses to anticipate demand for hundreds of millions of products every day — can be used to build precise forecasts for virtually any business condition, including product demand and sales, infrastructure requirements, energy needs, and staffing levels. It automatically provisions the necessary cloud infrastructure and processes data, building custom AI models hosted on AWS without requiring an ounce of machine learning experience on the part of developers….

Moreover, it says Forecasts’s predictions are up to 50% more precise than traditional methods…."  Read more   Hmmmm…. Whew… I was worried that Amazon had ruined life by being able to predict the future, but its just the old snake oil salesman’s "up to" trick.  He’s told us the ceiling… it can’t be better than "50% more precise" than"traditional methods".  Question is: how much less better or even worse can it be?  There may not even be an accuracy advantage.   Where’s the floor?  Minus infinity???    C’mon Amazon!! Where’s your accuracy metric?     Alain


Simply ClickBait


 Calendar of Upcoming Events:

September 4-6, 2019
Pocono Manor, PA


4th Annual Princeton SmartDrivingCar Summit

evening May 19 through May 21, 2020


On the More Technical Side

https://orfe.princeton.edu/~alaink/SmartDrivingCars/Papers/