15th edition of the 6th year of SmartDrivingCars
Saturday, March 31, 2018
The Most Important Self-Driving Car Announcement Yet
A. Madrigal, Mar 28, “On Tuesday, Waymo announced they’d purchase 20,000 sporty, electric self-driving vehicles from Jaguar for the company’s forthcoming ride-hailing service…. But the company embedded a much more significant milestone inside this supposed announcement about a fancy car. With orders now in for more than 20,000 of these vehicles and thousands of minivans that Chrysler announced earlier this year, Waymo will be capable of doing vast numbers of trips per day. They estimate that the Jaguar fleet alone will be capable of doing a million trips each day in 2020.
You could quibble with their math (will it really be that many daily trips per car?) or their overall utilization rate (how many cars will be lost to maintenance per day?), but if Waymo is even within 50 percent of that number in two years, the United States will have entered an entirely new phase in robotics and technology.
The company’s autonomous vehicles have driven 5 million miles since Alphabet began the program back in 2009. The first million miles took roughly six years. The next million took about a year. The third million took less than eight months. The fourth million took six months. And the fifth million took just under three months. Today, that suggests a rate on the order of 10,000 miles per day. If Waymo hits their marks, they’ll be driving at a rate that’s three orders of magnitude faster in 2020. We’re talking about covering each million miles in hours.
But the qualitative impact will be even bigger. Right now, maybe 10,000 or 20,000 people have ever ridden in a self-driving car, in any context. Far fewer have been in a vehicle that is truly absent a driver. Up to a million people could have that experience every day in 2020.
2020 is not some distant number. It’s hardly even a projection. By laying out this time line yesterday, Waymo is telling the world: Get ready, this is really happening. This is autonomous driving at scale, and not in five years or 10 years or 50 years, but in two years or less….” Read more Hmmmm…Yup!! This is HUGE! It will change the city and the key to making it so it doesn’t make thing worse is Ride-sharing. If we ride-share we’ll reduce energy, pollution & GHG by more than 50% and provide high-quality, affordable mobility indiscriminately for all. It becomes the new high-quality, low-cost mass transit. If it’s kept/operated as another alternative for the 1%ers to be chauffeured alone, then the outcome is UGLY. Ride-sharing is KEY! Alain
Smart Driving Cars Podcast Episode 32
F. Fishkin, Apr 2, “Waymo’s big partnership with Jaguar, a deadly Tesla autopilot crash and the plans for congestion pricing for vehicles in parts of Manhattan. In Episode 32 of the Smart Driving Cars Podcast, Princeton University’s Alain Kornhauser takes on the tough questions with straight answers and no hype, along with tech journalist Fred Fishkin. Listen in…and subscribe!“
Hmmmm…. Now you can just say “Alexa, play the Smart Driving Cars podcast!” . Ditto with Siri, and GooglePlay. Alain
Real information every week. Lively discussions with the people who are shaping the future of SmartDrivingCars. Want to become a sustaining sponsor and help us grow the SmartDrivingCars newsletter and podcast? Contact Alain Kornhauser at firstname.lastname@example.org! Alain
2nd Annual Princeton SmartDrivingCar Summit
Become a Sponsor and Display your wares
Waymo is buying up to 20,000 Jaguars and plans to rev up a driverless ride-hailing service
R. Mitchell, Mar 27, “In a dramatic commitment to driverless ride-hailing, Waymo said Tuesday that it is moving ahead with plans to put tens of thousands of driverless vehicles on public roads over the next several years.
Waymo, the driverless-technology arm of Google’s Alphabet, said it will buy as many as 20,000 luxury all-electric Jaguar I-Pace compact sport utility vehicles equipped with robot technology, and deploy them in its driverless fleet….Waymo said it will begin testing the I-Pace later this year. Already Waymo has “thousands” of hybrid Pacifica minivans on order, and will add the I-Pace to its commercial fleet in 2020 as it expands driverless ride-hailing to other cities. With 20,000 vehicles, Waymo said, it could handle 1 million rides a day….
“This is bad news for other auto manufacturers who may have been working to become Waymo’s [choice],” said Alain Kornhauser, who heads the driverless transportation program at Princeton University. “To serve the billion [car] trips that occur on a typical day in the U.S., they’d need about 35 million vehicles. That’s where all this is going.” There were more than 260 million cars and light trucks registered in the U.S. as of late 2016, according to IHS Market….” Read more Hmmmm…This is also bad news for all of the other OEMs that were hoping for a big order from Waymo and for startups that have been hoping that Waymo would make them the new “Otto“. Alain
Ambarella Demonstrates Its Fully Autonomous Vehicle on Silicon Valley Roads
Press Release, March 28, “Ambarella, Inc. (NASDAQ: AMBA) a leading developer of low-power, HD and Ultra HD video processing semiconductors, today announced it is demonstrating its fully autonomous EVA (Embedded Vehicle Autonomy) vehicle on Silicon Valley roads to industry analysts and customers. EVA has been trained to deal with the various traffic scenarios presented by Silicon Valley’s challenging urban environment. The fully autonomous car combines software and algorithms based on over 20 years of autonomous vehicle research with Ambarella’s low-power CV1 embedded computer vision processors based on its CVflow architecture. EVA’s high-resolution stereovision cameras deliver the 360-degree short and long distance viewing capability required for advanced perception and precise self-location. EVA includes sensor fusion of the vision information with Radar and map data to provide the information necessary for path planning and merging maneuvers without the need for additional LiDAR systems…” Read more Hmmmm…This is a substantial announcement from Alberto Broggi, one of the most experienced AV/ComputerVision innovators. This is one to watch closely because of its simple elegance. See also “Ambarella Introduces CV2 4K Computer Vision SoC with CVflow™ Architecture and Stereovision” Alain
What We Know About Last Week’s Accident
The Tesla Team, Mar 27, “We were deeply saddened to learn that the driver of a Model X vehicle involved in an accident last Friday passed away. Safety is at the core of everything we do and every decision we make, so the loss of a life in an accident involving a Tesla vehicle is difficult for all of us. Earlier this week, Tesla proactively reached out to the authorities to offer our assistance in investigating.
While we do not yet know what happened in the moments leading up to the accident, and we do not yet have any idea what caused it, here is what we do know:
Due to the extensive damage caused by the collision, we have not yet been able to retrieve the vehicle’s logs….
Our data shows that Tesla owners have driven this same stretch of highway with Autopilot engaged roughly 85,000 times since Autopilot was first rolled out in 2015 and roughly 20,000 times since just the beginning of the year, and there has never been an accident that we know of. There are over 200 successful Autopilot trips per day on this exact stretch of road.
The reason this crash was so severe is that the crash attenuator, a highway safety barrier which is designed to reduce the impact into a concrete lane divider, had either been removed or crushed in a prior accident without being replaced. The following image shows what the barrier looked like when the crash attenuator was in proper condition, and what it looked like the day prior to the crash, based on dash cam footage from a witness of the accident who commutes daily past this location. … …” Read more Hmmmm… Be sure to see the images. This information about how many uneventful traverses and the a catastrophic one emphasizes how important it is to recover the data from this crash and make it publicly available to everyone. Some unique combination of circumstance occurred in this crash that had not come together in the “85,000” traversals. This precise combination of factors needs to be identified and properly addressed by everyone involve with automated vehicle technology.
One thing that is lurking in each of these catastrophes, (Brown/Florida, Firetruck, Uber/Tempe and this.) is that the AEB doesn’t seem to be helping at all. Neither skid marks nor any suggestion of the application of brakes have been reported in any of these crashes. The common thread is a crash with a stationary object; one that has zero longitudinal velocity. We know that at least some Automated Emergency Braking (AEB) systems are designed to ignore the detection of objects ahead that are stationary. (It may simply be an overpass, a sign, an overarching branch, … that we can easily pass under, so don’t discomfort the driver by beginning to needlessly apply the brakes…. Good news: we disregard annoying false alarms. Bad news, we create false negatives with tragic consequences.). As I have written many times, we need to go back and substantially improve AEBs (Safe-driving aspects) so the false alarm rate is very small and we operate ‘Self-driving cars’ only in domains where ‘Safe-driving cars’ actually work and their collision avoidance systems are not turned off. Moreover, the concern needs to be at least as much about keeping our feet on the brake, as hands on the wheel.
While we are at it, I wish to also repeat my concern about Daimler’s implementation of Intelligent Cruise Control (ICC). Three years ago when I bought my MB S550 with Dystronic Plus, I noticed that a tap of the brakes would completely turn off the ICC. At the time I asked/complained about what human logic/AI system suggested that a tap of the brakes by me should be interpreted as “please turnoff the ICC”. I contend that the major safety function of ICC over conventional cruise control (CCC), is that it keeps me from rear-ending the car ahead should it begin to decelerate, or even come to a complete stop. CCC just controls the throttle and not the brake so it is of no help here. ICC was designed, and I was charged, for this extra safety & convenience functionality. But this wonderful feature is voided if the brakes are touched. The system is turned off and says “you are on your own, sucker”!
On the other hand, if I touch the gas pedal, the braking system is temporarily disabled so that it doesn’t work against my desire to go faster. Once I get off the throttle, the brakes are automatically available to work to keep me from rear-ending the car ahead. The ICC is NOT turned off.
I simply don’t understand why my touching of the brakes doesn’t operate is a similar way? Just turn off the the throttle throttle while I brake, or to be really safe, turn off the throttle completely until I actively turn it back on again. But DON’T also turn off the braking function so that I don’t rear-end the car ahead of me if I haven’t press hard enough on the brake, or, for heaven’s sake, taken my foot completely off the brake.
But no! Some designer/AI system somewhere in Daimler decided that “tapping brake” = “please turn off ICC”. Really??? (Or, did Daimler simply extend CCC which properly turns off the throttle function if the brake is touched, but since CCC has only one function, control the throttle, it turns off ICC and consequently turns off the throttle. Could it really be that Daimler, in its desire to turn off the throttle, simply replicated its CCC process of turning off the whole system, not realizing the potential consequences to the new functionality, braking?
In any event, I reported all of this to Daimler and NHTSA in 2014 and have yet to hear from either. Daimler doesn’t seem to believe in over-the-air updates and hasn’t made any attempt to upgrade my system during any of its maintenance cycles. I certainly hope that Daimler pays more attention to its other customers. I’ve only owned 9. It is unlikely that I’ll hit double digits. 🙁 Alain
Tesla confirms Autopilot was activated during fatal crash under investigation after reviewing data logs
F. Lambert, Mar 30, “…It’s still unclear how the vehicle moved from the lane to the barrier. ..
Here’s the blog post…
…In the US, there is one automotive fatality every 86 million miles across all vehicles from all manufacturers. For Tesla, there is one fatality, including known pedestrian fatalities, every 320 million miles in vehicles equipped with Autopilot hardware. If you are driving a Tesla equipped with Autopilot hardware, you are 3.7 times less likely to be involved in a fatal accident.
…” Read more Hmmmm… The bottom line is still very good, but this crashing into stationary objects needs to be fixed!. It starts with fixing the AEB. Alain
Musk’s Tesla faces a cash crunch as it confronts a suddenly skeptical Wall Street
R. Mitchell, Mar 31, “It was a bad week for Elon Musk. Tesla bonds got hammered, Tesla stock got slammed.
The National Transportation Safety Board launched an investigation into Tesla and its Autopilot self-drive system after a Model X driver was killed in a fiery crash on a Silicon Valley freeway. Late Friday, Tesla said Autopilot was engaged and the driver had his hands off the wheel for six seconds.
The company issued a voluntary recall for 123,000 Model S cars to check for problems on power-steering bolts.
Worse for the company’s financial future, bond-rating service Moody’s on Tuesday downgraded Tesla’s bonds. Ominously, the report said the $3.4 billion in cash and securities Tesla had at the end of 2017 was “not adequate to cover” its normal operations and production increases….
As of Friday, the Bloomberg Tesla Model 3 Tracker, which makes its estimates via vehicle identification numbers, put the weekly total at 1,076. Tesla might end up counting the figure differently…
Given the Moody’s downgrade and the generally higher cost of corporate high-risk bonds due to rising interest rates, bond analysts say Tesla would have to offer an interest rate significantly higher than the 5.3% in the last issue….
But what about the earthly need for capital and the recent hits to Tesla’s stocks and bonds? “The whole world is going to be screaming, ‘take my money, take my money,'” he said.
Even “Montana Skeptic,” a popular Tesla critic on the Seeking Alpha website, warned readers not to discount Musk’s financial resourcefulness….”Read more Hmmmm…Be sure to look at the Bloomberg report. It isn’t easy to really innovate. Alain
It’s the Edge Cases You’ve Got to Get Right
Ken Pyle, Mar 28, ““Data as the new source code,” was one of the many thought-provoking statements made by Nvidia’s CEO Jensen Huang in his keynote speech at Nvidia’s GPU Technology Conference. In the above interview, autonomous vehicle expert and Princeton professor, Alain Kornhauser, gives his take on Huang’s talk and other news, including: ” Read more Hmmmm…Be sure to watch the video. Thank you Ken. Alain
Fatal Tesla Crash Raises New Questions About Autopilot System
N. Boudette, Mar 31, “…For the company, the significance of the crash goes beyond Autopilot. Tesla is already reeling from a barrage of negative news. The value of its stock and bonds has plunged amid increasing concerns about how much cash it is using up and the repeated delays in the production of the Model 3, a battery-powered compact car that Mr. Musk is counting on to generate much-needed revenue.
It is also facing an investor lawsuit related to Tesla’s acquisition of SolarCity, a solar-panel maker where Mr. Musk was serving as chairman. Meanwhile, competition is mounting from other luxury car makers that have developed their own electric cars, while Waymo, the Google spinoff, General Motors and others seem to have passed Tesla in self-driving technology… Read more Hmmmm…Sorry, Tesla was never ahead of Waymo and, in some ways, they are still ahead of GM. Alain
Why This Chief Investment Officer Thinks Tesla ‘Is on the Verge of Bankruptcy’
Mar 28, ““Tesla, without any doubt, is on the verge of bankruptcy,” John Thompson, chief investment officer of Vilas Capital Management, wrote in a note to investors. Even as Tesla investors approved the largest compensation plan ever for CEO Elon Musk on Wednesday, the company has incurred nearly $5 billion in operating losses for shareholders. …As a reality check, Tesla is worth twice as much as Ford, yet Ford made 6 million cars last year at a $7.6 billion profit while Tesla made 100,000 cars at a $2 billion loss,” wrote Thompson, who has consistently outperformed the market as a hedge fund manager.
He said, “I have never seen anything so absurd in my career.”” Read more Hmmmm… I guess that’s why Waymo chose Jaguar over Tesla. Alain
Why it’s time for Uber to get out of the self-driving car business
T. Lee, Mar 27, “… Reaching commercial viability could easily require several more years of steady funding, and it’s not clear if Uber can provide that. In 2017, Uber lost more than $1 billion per quarter, and at its current burn rate the company will run out of money in 2019 if it can’t raise more. That’s a problem because it’s far from clear the company’s driverless car technology will be ready for a commercial launch by then. The tight financial situation could create pressures to rush Uber’s self-driving car technology into the marketplace before it’s ready….
More fundamentally, the project needs to earn back the public trust after last week’s fatal crash. That’s going to be hard to do if the project remains under the corporate umbrella of Uber, a company that doesn’t have the best reputation for honesty and respect for the rules. Selling the project to a new owner could give it a fresh start in the public mind. The most attractive suitors here are likely to be conventional car companies. They have significant financial resources and the staying power to fund the project for as long as it takes to get the technology right. More important, car companies generally have safety-oriented cultures with engineers used to coloring inside the lines. A car company would be good at making sure that the team is taking all appropriate safety measures—and also at convincing the public that the vehicles can be trusted….
The conventional Silicon Valley approach to technology development is what Facebook CEO Mark Zuckerberg used to call “move fast and break things.” A website can’t run people over, so it made perfect sense that Internet companies are used to introducing half-finished products into the marketplace and then iterating rapidly based on feedback from users. Obviously, this approach won’t work for self-driving cars. …
Meanwhile, emails unearthed during Uber’s recent lawsuit with Waymo revealed that Levandowski has been dismissive of safety concerns since his days as a senior engineer at Google. “We don’t need redundant brakes & steering or a fancy new car; we need better software,” wrote engineer Anthony Levandowski to Alphabet CEO Larry Page in January 2016. “To get to that better software faster we should deploy the first 1000 cars asap….”
The most obvious suitor here is Volvo…” Read more Hmmmm… The value of a driverless car is using it, not selling it. If you built geese that, without any effort, laid golden eggs you would only sell the geese for a price > (the value of the eggs + cost of building the geese). This makes it impossible for a “buyer” to be competitive with a “maker”.
Operating a fleet of driverless cars will be easier than managing a fleet of conventional Ubers. Thus the maker of Driverless can, “with less effort” deliver ride-hailing services. So if Uber wants to be competitive in the ride-hailing business they’ll need to “make”, because they won’t have a viable opportunity to “buy”, Driverless cars. Alain
As Uber death is investigated, its chief faces a crossroads on driverless tech
R. Mitchell, Mar 31, “When Dara Khosrowshahi took over as Uber’s chief executive last August, he faced a pile of festering problems. Near the top: The company’s troubled driverless-car program, then embroiled in a major lawsuit over trade-secrets theft with arch-foe Waymo, and suffering after a series of defections by top engineers.
He had three options: find outside partners for joint development of driverless tecNow, after an Uber driverless car killed an Arizona woman and threw the company’s program into turmoil, Khosrowshahi may be back at the same crossroads. But this time Uber’s options may be slimmer.hnology; drop the internal program altogether and buy the technology off the shelf; or continue to go it alone. Khosrowshahi chose the solo route.
The industry’s eagerness to separate itself from the fatality, on top of Uber’s well-earned reputation for over-aggressiveness and arrogance, could make partnerships more difficult to strike….
They may well be so far behind Waymo that they can’t catch up” with their own technology, said Alain Kornhauser,…
“I don’t have the sense that [Uber investor] SoftBank bought into Uber because of their driverless vehicles,” Tusk said. “If ultimately investor interest is around ride-sharing itself, maybe Uber doesn’t need to make this the focus.” Read more Hmmmm… My view is that SoftBank invested because they expect/want Uber to expand at least 10x. Uber might be able to manage 2M gig workers, it is doubtful that Uber, or anyone, can manage 20+M workers. That is way too many “cats”. Orchestrating 20+M driverless cars… No Problem!!! So SoftBank did buy Uber because of its driverless vehicles or SoftBank is intending to develop or has its own driverless technology. SoftBank’s chance of “getting” Waymo’s technology is “Slim2None”. Alain
Uber’s use of fewer safety sensors prompts questions after Arizona crash
H. Somerville, Mar 27, “When Uber decided in 2016 to retire its fleet of self-driving Ford Fusion cars in favor of Volvo sport utility vehicles, it also chose to scale back on one notable piece of technology: the safety sensors used to detect objects in the road. …
In scaling back to a single lidar on the Volvo, Uber introduced a blind zone around the perimeter of the SUV that cannot fully detect pedestrians, according to interviews with former employees and Raj Rajkumar, the head of Carnegie Mellon University’s transportation center who has been working on self-driving technology for over a decade….
Uber referred questions on the blind spot to Velodyne. Velodyne acknowledged that with the rooftop lidar there is a roughly three meter blind spot around a vehicle, saying that more sensors are necessary….” Read more Hmmmm…At this point, I still disagree. Even if there is a blind spot of 3 meters around the vehicle, this crash was not about 3 meters, but the recognition that a pedestrian was crossing the road 50+ meters ahead and was within a very narrow (~10 degree) field of view. Turning off of Volvo’s AEB/pedestrian detection might have actually been a good thing since its pedestrian detection system is effective only if the car is going between 2 and 30 mph ( or at least it was that way in 2016). So Uber needs to come clean and release all of the data used by the car in its AV algorithm(s) in the last 4 seconds before this crash . It also must list precisely what decisions its AV algorithm(s) made with respect to object detection, identification and throttle, brake, steering commands. Alain
The Self-Driving Car Industry’s Biggest Turning Point Yet
K. Roose, Mar 29, “One perk — or hazard, I suppose — of being a technology writer for the past few years has been getting invited to ride in a bunch of autonomous vehicles. I’ve been shuttled around in nearly a dozen self-driving prototypes, including a Ford in Michigan, an Uber in Pennsylvania and a Chrysler minivan in the California desert Whenever anyone asks what it’s like to ride in self-driving cars, my reply is: “Which self-driving cars?” Casual observers tend to talk about the progress of autonomous vehicles as if they’re a homogeneous category, but there is an entire spectrum of capability and safety. I’ve had calm and boring drives, and terrifying white-knuckle trips. The best results aren’t always from the companies with the most money or the most sophisticated marketing campaigns.
This week, the self-driving car industry reached a turning point, one at which the mixed progress of self-driving car projects became very clear. There was continued fallout from the fatal accident this month involving a self-driving Uber vehicle in Arizona, including a blockbuster story from my colleague Dai Wakabayashi about Uber’s struggles to get its self-driving cars ready for the market. Since the crash, Arizona officials have ordered Uber to suspend its autonomous vehicle testing program in the state, and Uber is not renewing its permit to test self-driving cars in California. In addition, Lior Ron, an executive who led Uber Freight and was involved in the company’s early self-driving efforts, is leaving the company….” Read more Hmmmm… Nice weeks summary. Alain
The driverless car revolution must proceed with caution
D. Norman, Mar 24, “Imperfect automation, continually getting better? Or distracted drivers, continually getting worse? Choose.
Recently, one of Uber’s autonomous automobiles was involved in an accident where a pedestrian was killed. What lesson should we learn from this incident? During the three years that my colleagues and I have been doing research on self-driving cars, this is the first death. Compare this single death with the 120,000 people who have been killed in automobile accidents in the United States in that same period: roughly 100 people each day. Fully autonomous cars have driven around four million miles rather than the nearly nine trillion miles driven by American drivers in that same period. The accident record is impressively low: in four million miles of driving, one death compared to 40 deaths in regular driving….
The potential for autonomous vehicles to produce tremendous saving of lives and injuries while increasing our quality of life provides strong support for the eventual introduction of fully automated vehicles. Nonetheless, just as new medicines and medical devices enhance lives, but their introduction is done cautiously, with carefully controlled tests, we must do the same with our autonomous vehicles. I look forward to the day when my self driving car will free me from the tedium and danger of driving. But that day is not yet here.” Read more Hmmmm… Don, very nice. Alain
On the Uber Autonomous Car Accident
T. Lassa, Mar 30, “Auto journalists and other various critics weighing in on last week’s tragic Uber driverless car accident have split into two factions. Those of us who never want to see robotic cars take over…Those of us who either consider the rapidly developing technology a good thing for safety overall, or at least an inevitable thing, ….” Read more Hmmmm… Shallow views from conventional Automobile media. Alain
Some other thoughts that deserve your attention
Congestion Pricing Falters in New York, Again
W. Hu, Mar 31, “Governor Andrew M. Cuomo set the stage for an ambitious congestion pricing plan when he declared that it was “an idea whose time has come.” But that time is not now.
There was little about congestion pricing in the state budget negotiated Friday by Mr. Cuomo and state lawmakers despite months of lobbying by advocates, a six-figure media campaign, and rallies by transit riders. The most significant development was a new surcharge that will be tacked on to every ride in for-hire vehicles in Manhattan south of 96th Street: $2.50 for yellow taxis; $2.75 for other for-hire vehicles, including Ubers and Lyfts; and 75 cents for car pool rides such as Via and UberPool. … ” Read more Hmmmm… Why are Yellow taxis favored over other for-hire??? And why are car-pool rides charged anything when they are arguably more efficient than the buses on the streets. Excellent that the charges are only in Manhattan below 96th where transit service is excellent and not North of 96th and the other boroughs where, for many trips, transit service leaves a lot to be desired. The fact that private single-occupancy cars are not being “charged” is very unfortunate but the parking industry must have exerted some political pressure. Parking charges are the de facto congestion charge in Manhattan.
Also, why is this called “congestion pricing” which has negative connotations? Why isn’t it called “value pricing”? It delivers “value” to those that choose to pay!! Alain
Half-baked stuff that probably doesn’t deserve your time
Enough With the Trolley Problem
I. Bogost, Mar 30, “You know the drill by now: A runaway trolley is careening down a track. There are five workers ahead, sure to be killed if the trolley reaches them….” Read more Hmmmm…For all of you that have nothing better to do in making these systems not crash into Firetrucks, barriers and pedestrians, you can worry about simultaneous events when essentially no two rare circumstances happen simultaneously at the same place because of the precision in the time dimension. Alain
Calendar of Upcoming Events:
2nd Annual Princeton SmartDrivingCar Summit
May 16 & 17, 2018
Save the Date
Application NOW OPEN
Become a Sponsor and Display your wares
On the More Technical Side
Capsule Networks (CapsNets) – Tutorial