1. An executive order is NOT a law. The next governor can cancel it at whim. The legislature, by passing a law, can change it.
2. An executive order, as we saw at the federal level with both President Obama and President Bush before him - with the authority to use broad administrative powers - can be a force for funding and direction. But see #3.
3. Every state is different. The governance structure - who has the power and authority - varies greatly from state to state, from legislature to legislature. In some states, the governor is the power center; in others it is three top legislators who stay forever and do not need to listen to anyone else.
Washington State does more than grow apples and technology ... it wants autonomous vehicles
I claim total ignorance of the power structure of Washington State politics, but with so much technology business, the Seattle area cannot be happy that Silicon Valley in California is synonymous with the current tech boom. The governor is lagging behind, but paying attention. Who knows what he tried to get going before deciding on an AV executive order.
[Photo from HUD.]
Lots of Whereas points before the meat (or vegan meat)
Likely following state tradition that an executive order must establish that a bona fide problem exists before one should be issued, the Washington State executive order begins with a list of "Whereas ... " there's a problem and opportunities here.
[Photo by from AJAD MOHAMAD (OWN WORK) [CC BY-SA 4.0] from livability.com.]
Problems identified include:
Lots of auto "accidents" - not PC; the correct term is crashes,
The state wants to be in the technological lead - it's already behind in that race for this industry.
Low and behold, the executive order has found the solutions of (a) safety and (b) economic development in autonomous vehicles.
The nitty gritty
So how is the magical solution of autonomous vehicles going to happen in the land of apples and Microsoft? A study group - ta da! Actually, it's being called a "work group."
Not kidding. Obvious that no one on the governor's team is from the world of marketing.
But wait ...
The one promising point in the executive order is permission for pilot programs that allow for autonomous vehicles to be tested on Washington States roads. The vehicles must have on board equipment that can be operated by a human and an actual human physically present UNLESS (among other predictable details):
The vehicle is "capable of bringing the vehicle to a safe condition in the event of a system
failure,"
Developers of the vehicles self-certify to DOL (Department of Licensing) that they are compliant with the above
requirements.
No requirement of a data recorder, no requirement of coordination with local jurisdictions, and no extraneous requirements.
I make fun, but this executive order is pretty middle of the road in terms of state policies and laws.
[Photo: US Transportation Secretary Anthony Foxx (R) and Google Chairman Eric Schmidt (L) ride in a Google self-driving car at the Google headquarters on February 2, 2015 (Photo: Getty) via inews Essential Daily Briefing]
Crushing news for the cuteness factor
Say it ain't so. Did Waymo not learn the Toyota Prius v. Honda whatever lesson that conspicuous upgrades or do-gooding or technology quickly becomes a desired wow factor? Nope. Waymo is ditching the cute, pod-like driverless vehicles that Google developed and tested. Ah, I remember fondly the artistic designs put on the vehicles and the sweet factor.
Sorry, there's no charm in the look of an average car. A little sad here that the fireflies, as Google named the two-seater podcars, are disappearing. According to the inews Essential Daily Briefing:
Firefly will make a trip to the Arizona Science Centre in Arizona during August, before making its final journey back to Texas in October to commemorate the second anniversary of the first autonomous drive. They are also available to view in the Computer History Museum in California, and in London’s Design Museum.
Better yet - not - Waymo is counting on a minivan, which is inaccessible for people with disabilities and has zero charm, to catapult it to greatness with families who lug gear.
Today's post is brought to you by the acronym of the day, CBI. Since DC is a town of acronyms, here is one that does not stand for the name of an agency, but instead means confidential business information. This is a post about yesterday's public meeting held by NHTSA. I do not believe a video of the event will be made available because there was not a live feed. So much for transparency.
Good outreach does not equal posting one announcement on one government webpage, okay?
Here's my take on yesterday's NHTSA public meeting and it begins before the meeting took place. First, before getting to the meeting, let me say that the outreach was terrible and limited. I'm sorry, but an official announcement with just a few days open for registering to attend the meeting - which was not even held at the US Department of Transportation building - displays an impression that you only want people in the know, an inner circle, to attend. In the age of Twitter, Facebook, and tons of social media outlets, that is poor and inadequate outreach and public engagement. Second, my bad, I am ditzy and inattentive (which is why I really want driverless transportation), but I failed to read the fine print that the event was being held at the GSA (General Services Administration) building, which is in an entirely different part of town than DOT. Fortunately, both agencies are only a couple of Metro stops away from each other on the same Metro line.
I attended even though mid-level NHTSA employees hosted the meeting at an obscure location. The DOT building is quite large, so I am not sure why the meeting was not held there.
It is so weird for me to be in a room with the likes of Google, Ford, Toyota, GM, and Zoox. The one interesting misfit, besides me, was the guy (did not catch his name) from the Commercial Vehicle Training Association (CVTA), a group doing its best to keep drivers at the wheel and attending their member driving schools. Unlike the American Trucking Associations (that is a singular entity, by the way), which has a big interest in cheap, efficient trucking operations sans drivers, CVTA is a voice for delay and roadblocks to thwart a driverless trucking industry.
Points of agreement from those actually in the no-decor government auditorium
Except for the CVTA guy, who has diametrically opposed fish to fry than anyone else who was in the room (a truly boring, small auditorium with no windows or interesting decor), everyone - tech and car companies and public advocates alike - wants clarity and safety. Most want flexibility to support innovation. I say most because I do not get that feeling from one public safety organization, Consumer Watchdog, which has one particular representative at every NHTSA driverless event.
In terms of public advocacy groups speaking out there was Consumers Union, which is the policy and action division of Consumer Reports; the Future of Privacy Forum, which advances "principled data practices" and self-regulation (more on that oxymoron later) because the pace of technology development is inconsistent with the speed and constraints on the process of regulatory change; and Advocates for Highway and Auto Safety, which has worked for decades on auto safety issues.
Car companies with representatives who spoke were Ford, General Motors, Toyota, and Hyundai. Tech companies with representatives who made statements were Google, Zoox, Lyft, and Bosch. Industry associations and representatives were the Center for Automotive Research (the one housed at Stanford) and the Association of Global Automakers. An interesting missing person was David Strickland of the Self-Driving Coalition for Safer Streets, but I came late due to other obligations, so it is possible that I missed him as well as representatives for people with disabilities, who have been active and enthusiastic driverless proponents.
The public comment period is open until Nov. 22 and every single representative who spoke yesterday stated that his or her company or organization will be submitting written comments. To submit comments in response to the proposed NHTSA guidelines, officially known as the Federal Automated Vehicles Policy, visit the Federal Rulemaking Portal (sounds very Harry Potter) and type in NHTSA-2016-0090. Or go straight to this comment page for the docket number. Much easier.
Where's that acronym already? (Say that with a Bernie Sanders voice.)
I will refrain from sharing my notes and instead concentrate on writing a quick summary about agreement and disagreement in the room, while pointing out what I find ridiculous. Being an outsider gives me a possibly less informed, but, perhaps, less biased and bureaucratic perspective.
A major concern of both the tech and the car industries is the protection of - here is today's acronym - CBI, or confidential business information. No one wants their trade secrets, more touchingly known as intellectual property, to become common knowledge. But an acronym is great because you have to be in the know to get it, though this one was easy to figure out. I was not the only one in the room who had not heard that particular acronym before, according to public remarks made.
Agreement List
I'm barely mentioning and not even listing this: Everyone except the CVTA guy loves driverless. What produces agreement and disagreement are the many details. It was a driverless, bureaucratic, mostly dark-suited love fest.
1. Preventing CBI from being disclosed. Every player producing, developing, or hoping to make tons of money on driverless vehicles wants to protect the CBI of themselves and their partners. Knowledge is power and no one wants to share a potentially superior and cheaper algorithm, sensor, AI software (okay, my tech expertise is non-existent). I did just watch the Imitation Game last night and I'm thinking of AI as a kind of Turing machine.
The balance of public disclosure versus CBI: Though there was, at the very least, lip service paid to transparency and the idea of information available to the public, there was broad agreement for the idea that as much information as possible be made public. That said, once details were touched upon, corporate interests expressed discomfort with any intellectual property - CBI- being open to scrutiny.
2. Trigger for new assessment letter or update - Should be a significant change, such as a change in the level of automation. There's not complete unanimity, but broad agreement. Otherwise the regulatory burden will become unbearable.
3. A little less agreement, though there was some consensus that for purposes of regulation and paperwork there should be a point person or company. Where several companies contribute equipment, software, and hardware to a driverless vehicle, who answers to and must file documents with a regulatory agency? You see where this gets into the weeds very quickly. The consensus was to have one entity per vehicle type (that's my language) be responsible so that every vehicle does not require massive paperwork from many companies. OMG, there will not be sufficient agency staff to do an adequate job in that reality.
4. The Safety Assessment Letter - get ready for it, because a new acronym is about to hit your brain, SAL - should have a cover or summary page up front. Basic contact and other information, probably a summary, should be on page one. Now while the details are not clear, the concept attracted a broad consensus.
5. Government should be supportive of corporate innovation, flexibility, and driverless in general.
6. Clarity, clarity, clarity on everything, but specifically the scope of the guidelines, and timing of required documentation to be submitted.
7. Quick response to SAL that it is complete or incomplete - and, if incomplete, how so. GM suggested that a response come from NHTSA's general counsel. Suggestions of 30 days were made, but there were others.
8. Template or templates for SALs were requested by several speakers. This will help both the parties preparing these documents and those analyzing them, whether inside or outside of government.
Disagreement and open questions list
1. Self-regulation or actual enforceable, mandatory regulations? This breaks down along predictable lines of tech and auto industry players versus safety advocates. With a new and completely different kind administration coming to Washington so soon, NHTSA under the Obama Administration will lose its chance to answer this question. The hearing provided an opportunity for public comment, but this game has just begun.
2. Triggers: Trigger level for regulation as an autonomous vehicle and for safety assessment letter: Is SAE level 2 the trigger? Level 3? Are the levels in practice actually clearly distinct? There are lots of opinions on this issue.
In fact, the whole issue of what constitutes a significant update triggering an update of an SAL is far from clear. Very in the weeds and I await a spectrum of comments from those far more conversant with automotive policy and government regulation. And may I say that this is where the average person needs experienced and knowledgeable players in government because the gobbledegook of regulatory details is a foreign language that one must be learned about to participate effectively.
3. Testing versus vehicles sold or in operation: There was disagreement or at least divergent opinions expressed about how testing should be regulated, particularly testing on public roads - like next to your car - versus how to regulate an industry that will be selling, leasing, and likely providing the contemporary equivalents of taxi, jitney, and transit service.
I thought it was interesting the mentions of consumer information because they harken to the 20th century model of individually-owned cars and light trucks versus the distinct possibility for driverless of substantial shared-use services, particularly in dense urban areas and suburbs. Does anyone look at vehicle or airplane safety information when they get into an Uber or onto a plane? That's not happening.
4. Timing of required filing of SAL - using that acronym already. Should the filing deadline be three or four months before an update or new equipment is put on the road? More? One suggestion was made of a mandated annual update for each letter. That was Lyft, but the word "mandate" is mine and perhaps I am misreading the statement made.
Singular points raised
Since industries sometimes cover up health and safety risks - not only the tobacco and football powerhouses - it is relevant whether there will be whistleblower protection and a safe space for companies to reveal unsafe conditions and things that need to be fixed. This raised the issue of tort liability in case of accidents, which is a deterrent to releasing information that admits of or suggests imperfect hardware, software, and equipment. A fear raised at the hearing was that safety data mandated for disclosure to NHTSA (or, presumably, to state regulators) would set companies up for liability in case of accidents.
Me to industry - Just take the lead of Volvo and the few others that are taking my advice, or, more likely, coming to the conclusion on their own to take the airline industry model of effective strict liability and promoting a safety culture. Pay out some money and forget about the who is actually liable. There will be accidents and the public appreciates assumption of responsibility rather than obfuscation. We all hope that driverless will be way safer than human-operated vehicles, but nothing is completely safe.
The Future of Privacy Forum (FPF) speaker was interesting for advocacy for protections of privacy, but not for a government mandate to protect personal data, or, in privacy-speak, requirements for deidentification of data. FPF dreams of a world where companies voluntarily adhere to standards, which effectively are agreed-upon, crowdsourced regulation or industry-wide standard practices, all embraced by the public. I'm on the fence.
nuTonomy - One of the driverless cars tooling around Singapore in anticipation of AV-taxibot services got into an accident with a truck. In marked contrast to Tesla, but in taking a page from Google PR on handling minor accidents, nuTonomy has been up front about what happened: nuTonomy has halted driverless operations and it has released all data. Nvidia - This AI savvy company is cuddling up with Tesla, which makes sense because the founder is the proud owner of a Tesla car. Musk and Huang, Nvidia's CEO, have met at various forums and now the companies of these friends are going to capitalize on Nvidia's AI system for driverless vehicles, DRIVE PX 2, which will be used in Tesla cars.
DRIVE PX 2 is an end-to-end AI computing system that uses groundbreaking approaches in deep learning to perceive and understand the car’s surroundings… [and] lets carmakers first train their own deep neural networks on GPU supercomputers. Once loaded into the car, it processes the networks at high speed to provide the real-time, accurate response required for autonomous driving.
It is safe to say that Nvidia is the leading or one of the leading companies developing AI for driverless vehicles. Nashville - The Tennessee city has been selected as one of five international cities for technical assistance in planning and technology to prepare for driverless and connected vehicles. The other cities are Austin, Paris, Buenos Aires, and Los Angeles; plus, five more will be added within the next two months. Today is for the letter N, so we will celebrate Nashville. The program is officially called the Bloomberg Aspen Initiative on Cities and Autonomous Vehicles and it aims to develop a set of policy recommendations for ways cities should plan and use their land and resources for when automated vehicles arrive on their roadways. With a word from the letter W Wheego is changing its image and its purpose from being a tiny electric car company to being an autonomous car company - yet another driverless testing permit for the state of California - with an emphasis on deep learning and artificial intelligence (AI). Using the word of the month, pivoting is what Wheego is doing and the pivot has a geographic element. The company is shifting its focus from the US to China. The new official name of Wheego is Wheego Technologies. Alphabet bomb Google is ready for its car business to go out on its own. Google's Alphabet car will soon be a standalone business. Corporate legal stuff (a technical term) is in the works. No plans released yet for how the cars or travel will be sold. Grammar Comma.ai has been stopped in its tracks. Slated to sell its driverless-in-a-box add-on for existing cars by the end of this year (within the next two months), NHTSA has warned Comma.ai to not sell its product due to possible conflicts with federal regulations.
If you follow driverless news each day, you already know that Google invented something for its software that alerts its self-driving cute cars to yield to emergency vehicles. Here's a random article. There's also Volvo, which promoted with fanfare its assembly line roll-out of its driverless cars. Swedish drivers will be testing those. New York drivers would be better, but I say go for the lunatics on this one. Back to Google: Is it only me who thinks this, someone who does not have a clue how to develop the simplest software and never looks under the hood of a car, or are the AI (artificial intelligence) folks are already past this one-small-situation-one-bandaid approach to driverless software? That's the whole point of AI isn't it? You teach it through accumulated experience. By way of analogy, by the time your kid is learning to drive, he or she does not need to be taught specifically what to do in every little situation. Part of what we want from driverless is what we want from humans - judgment in unusual circumstances. We also want the capacity to pay close attention in endlessly boring situations, something humans are not good at. Where's the Google ride? So I ask if Google has slipped? Why weren't they before Uber or all of these transit and taxipod pilots in introducing their cute pod cars to the general public? Has Google become a prudent, risk-averse, large, conventional corporation where innovation gets lip service but not real support? I don't know. I'd have to ask the 20-somethings I know who work there. Speaking of Uber, okay, yes, I'll be the 50th or 100th site to post one of the videos of giddy journalists who trekked to Pittsburgh taking a driverless Uber ride. Well, with a driver and an engineer and lots of explaining. I'm sure there's a million of these videos already. Here's the one from Wired. It's cute, but not on youtube. Relax in the self-driving Uber backseat Here's a six-minute video from the PBS NewsHour. (Yes, even years later, I still think of the show as the MacNeil-Lehrer Report. MacNeil co-wrote an entertaining history of the English language, by the way. Quite readable. There's a companion series that aired on, where else, PBS.)
Already with ride hailing (Uber, Lyft, etc.), micro-transit (Chariot, Brid), shared rides (UberPool, LyftLine, Via, Split), and even bikeshare, our larger cities beyond the five boroughs are changing into places where one does not need a car, or where going car-lite is entirely feasible. But outside of the relatively few places where transit works well - usually with rail, a few with BRT (bus rapid transit) - will the driverless revolution be the last nail in the coffin for transit viability in places large and small where there is, using a technical term here, crappy transit? Ford thinks so; the company is buying Chariot, a San Francisco micro-transit operation.
Toyota is investing in the University of Michigan (sorry Buckeyes) to the tune of $22 million for artificial intelligence research that will go into driverless vehicles and many other types of products. The Japanese company is also interested in robotics products that enable older adults to remain independent and to assist people with disabilities. In a move that I perceive as related to autonomous vehicles, but is not technically so, Google has partnered with Israeli company Gett, a ride hailing, app-based, business similar to Uber. (Now I can't help but mention that according to Jewish law a gett is a document related to divorce that a husband must give to the wife - but without any reciprocal requirement - if she is to be free of the marriage.) Cozy linkage The Jerusalem Post reports that
In the deal, Google will offer information and a link to Gett in its popular Maps app when users search for directions, including an estimated fare and travel time. The feature, which previously went live in the UK and as of Wednesday is available in Israel and Russia, lets users seamlessly jump from Maps to Gett to order their cabs, or directs them to the app store to download the app. Google has offered the same feature with Uber for some time ...
The Gett deal is just one of a number of similar deals between Google and ride hailing companies around the globe. The deals demonstrate that Google perceives the ride hailing market as significant. Or perhaps Google is hedging its bets because no one can be sure of the business models for future transportation. These deals put Google into the space that RideScout formerly inhabited, though that app company did not operate on a global scale. In the musical chairs game of transportation app businesses, Moovel, which is owned by Daimler, purchased RideScout in April 2016. From the warehouse to the street? For those who doubt that driverless vehicles are another few decades away or that they will not play nicely with walking humans (otherwise known as pedestrians), this excellent article about the use of self-driving forklifts being employed in factories might give pause. The article does a good job of explaining the technology and operations. Data convinced union leaders to allow these (or not to oppose them). Experience demonstrates that the autonomous forklifts are safer than human-driven ones and that there have been no collisions with walking human factory workers. Oh and they save the factories lots of money as well.
Chris Ursom has left Google and he either does not know or has decided not to announce where he is going. Others have left, but Urmson has been the public face of the Google car project in the media and before government agencies. He had seemed to completely identify with the project. I can't imagine he will abandon the driverless world or that no other AV player will hire him. Forget Google, drive your own Test out your own driverless contraption now without being part of a tech company or auto company team. A driverless test track opens to normal people, but only on limited days in a roaming series of continent-hopping events. Applications seem to be necessary; you can't just show up. And there's a fee. So far, the events, well into 2017, will take place in the UK and in Europe. I feel that post-Brexit we can be honest and say the UK and Europe are different continents.
Kendall Square challenges Silicon Valley A partnership between MIT and the Defense Advanced Research Projects Agency (DARPA), an agency for innovation within the Department of Defense (the Pentagon for labeling purposes on this blog) is bearing fruit with the creation of a lidar chip that is super tiny and super thin and - best yet - possibly super cheap at $10, assuming mass production in the future. The chip is undergoing further adaptions to increase its power and accuracy. Honestly, I do not understand the technical details, but I do comprehend that this technology could make driverless technology safer and cheaper, therefore closer to being possible for mass production and use. Israeli startup also goes for less expensive LIDAR Not as cheap as the $10 goal, but Innoviz, an Israeli startup with millions in funding (okay, not unusual, but it's $9 million) also sees the holy grail of budget-happy LIDAR, though with an unspecified under-$100 price tag. This from the same Geektime article linked above. ccc
Innoviz is building LiDAR sensing systems, which are often used by satellites to measure distance and terrain. What Innoviz and some other companies are betting on is not only the growth of the autonomous vehicle market, but the inevitable reliance of such systems on LiDAR. Their main product, High Definition Solid State LiDAR (HD-SSL), won’t debut until December. They are claiming it will have a wider field of view than the current standard, “higher resolution in both axis and long range sensing,” and be cheaper at a cost coming in under $100.
When I was a little girl, every Rosh Hashanah (Jewish New Year) we would pass a car dealership, with new cars, on our walk to Sheepshead Bay, that's the real Brooklyn, to throw away our sins. It's true. That's why the bay needed to be cleaned up. But we could not pass that dealership without a big pause. My father loved to look at those new, shiny cars in the window. Well, we are sufficiently into the driverless age that we are getting second and third generation prototypes. Honda Acura - Seems fancy. No technology breakthroughs; no design wows. This is NOT completely autonomous, which Honda is saying is at least 15 years away. Translation: Honda is nowhere near ready. Here's a link to some more details and a video. GM - Chevy Volt EV, with driverless sensors, spotted on the streets of San Francisco. We need Karl Malden and Michael Douglas to investigate. (If you are a millennial, you probably do not get the reference to the long-ago TV show, the Streets of San Francisco.) Here's the theme song.
Google - According to several reports (I'm linking to just one), Google has patented and is, presumably, adding to its driverless vehicles a system to be used when a car crashes for spreading or oozing a sticky substance, akin to flypaper, onto the exterior of the vehicle. Wait, this is brilliant. Meant as a pedestrian safety technique, the flypaper-like substance would stick to the hit pedestrian - gluing the person to the car - so that the pedestrian would not be flung and further injured. Messy, but not harmed as badly. On all normal days, the car would have an "eggshell like" covering so that ordinary leaning against a vehicle would not result in ruined clothes and sticky hands.
I'm just renumbering here and giving links. So, here's comment number one for today. Comment #1: From SAE International, which has been doing research for DOT, the comment requests that NHTSA employ standard terms and definitions so that a plethora of confusing terms do not blossom like a garden of weeds after a summer rain. Okay, the comment did not use such language. Comment #2: This comment is from the Future of Privacy Forum and, as the name suggests, privacy of data is the organization's concern. This is the worry in a nutshell.
... [T]here is a risk that information like
pickup and dropoff locations may allow government agencies—or anyone else who
obtains this information through FOIA and other mechanisms—to identify individual
riders by associating it with publicly available records.
Comment #3: Ford Motor Company, which sent a representative to both NHTSA hearings, submitted this comment. Safety? Data security? Data recorders in case of accidents? Ford is in favor. I think it votes in favor of motherhood and apple pie as well. It's an educated guess. Oh and the company says that it is looking forward to working with NHTSA for safe autonomous vehicles. Comment #4: AAA submitted comments that raise a few issues. Here's two: data privacy and driver/operator training to use driverless vehicles. Okay, in my mind, the whole point of driverless is that I don't have to do anything but tell the vehicle where to go. I'm still dreaming of the tiny house that's driverless. Comment #5: Google's comment discusses succinctly an important concern that 50 states with 50 laws and 50 different sets of definitions and requirements would be a nightmare - my word. In a nutshell, this is Google's point.
We strongly urge that the guidance to the states recommend that
those states intending to regulate the deployment and operation of automated vehicle
technologies adopt NHTSA's operational guidelines with respect to the safety of the vehicles
rather than impose different, conflicting state rules. We have seen 15 states propose such laws
over the last 12 months; all of which have different scope and different definitions.
Comment #6: More from Google. The company comes out in favor of manufacturer self-certification that safety guidelines are met. Wow. Google is asking for a lot of trust and, let's face it, corporations have not always deserved that trust. They have not always looked out for the public good or even the good of their actual consumers. Google talks about operational limits of vehicles on the way to full automation in all types of places and in all types of conditions. Add to that, Google's point that the vehicle must reliably recognize those limits. Has Google seen the Nvidia video? One thing very heartening in Google's comment is the commitment to helping people with disabilities who are currently unable to operate our current driver-required cars. In the interest of my own time, here's a list of issues/thoughts in Google's comment. 1. Data recording so that we learn from crashes and other performance errors. Google goes into the fact that for current vehicles, not all crashes are reported - only ones in which the police are involved. Google would like to see collection of broader sets of data and common standards that allow for comparison of data. (I hope I am reading that correctly.) 2. Fully autonomous vehicles should have good crash avoidance technology. 3. Full scope of testing and data privacy protections. 4. Encouragement that NHTSA allow for advances in sensor technology.
We therefore
recommend any guidelines proposed by NHTSA should avoid focusing on being prescriptive
about equipment and instead focus on overall performance. The functionality and reliability of
components such as sensors should be examined as part of the vehicle manufacturer’s
functional safety analysis as it pertains to the system or subsystem in which the component is
used.
5. Google urging that NHTSA somehow block a patchwork of state laws. 6. Google - and me - wants permission for operation of vehicles without drivers and without driving equipment. 7. Google warns against closing off of innovation relating to technology to avoid crashes involving pedestrians - and, I presume, bikers, deer, etc. 8. Google foresees easier repairs in cases of recalls. The company wants to continue with existing regulatory standard. 9. Google wants NHTSA action to recognize value of computer simulations and modeling. 10. Google includes an appendix with scenario types used by its self-driving vehicle team. Comment #7: This comment comes from the Alliance of Automobile Manufacturers, meaning the old car companies, representing 77 percent of car and light truck sales. The crux of AAM's three-page comment is:
Facilitation of the deployment and adoption of Advanced Driver Assist Systems (ADAS), active safety
systems and eventually driving automation technology starting with a nationally-applicable legal and
regulatory framework that avoids a patchwork of state laws and regulations and maintains current selfcertification
practices, will accelerate the availability of such technologies and achievement of the
potential benefits.
Everyone wants to avoid the nightmare of 50 different states with 50 different regulatory regimes, standards, and sets of definitions. Comment #8: Delphi Automotive shares its "proven expertise" with NHTSA. That's Delphi's words - not that it's wrong, just bold. Delphi envisions a future that has both autonomous and connected vehicle technology. It's comments address both as part and parcel of future vehicles. I'll number these as well. These companies are verbose. No judgment; I just don't have much time. This long comment goes way into the weeds - important, geeky weeds. 1. Definitions are important. 2. Operational boundaries and detection of those boundaries, which will all change over time as the technology becomes able to operate in more places and situations. The comment also covers the issue of partial automation and driver takeover. I love that Delphi mentions that it can take up to 40 seconds for a driver to takeover, even mentioning that a driver might be asleep when a takeover request is made. Thank you Delphi for raising this important consideration. 3. Delphi diplomatically suggests communication with and involvement with groups representing people with disabilities as we move toward a driverless future. 4. Delphi calls for archiving of data as the learning and development process happens - and thereafter. (Does data ever disappear?) 5. Delphi wants NHTSA timeline for crash avoidance to "not slip." Current deadline of 2016. 6. Yes to electronic and cyber safety. 7. List of some technical standards. 8. Nice discussion related to pedestrian safety and signage/signals for AV. 9. Need for recognition of emergency vehicles. 10. Discussion of recalls and in what circumstances AV should be permitted to continue to operate. Comment #9: Orbit City Lab posits the "what if a terrorist/evildoer thing happens?" Comment #10: Also from Orbit City Lab on the terrorism issue. Comment #11: Comment from the Association of Global Automakers, meaning car companies outside the US. Here's the list of wants. 1. Again, that word "patchwork." No one wants it when it comes to different state laws with different standards and conduct permitted or prohibited. 2. AGA wants standard definitions. 3. Manufacturer flexibility. Not wanting to be hampered by slow regulatory approval processes. 4. Need to incorporate connected vehicle technology issues. 5. International harmony. Standard standards and singing kumbaya. Comment #12: Oh gee, another Orbit Lab paranoid comment. Comment #13: Ditto.
Michigan and Texas are investing heavily in driverless research facilities and working to lure private companies with big bucks. Driverless advancement is not the only venture contemplated. Robotics and artificial intelligence are also on those lists. Toyota is going to Ann Arbor to team up with the University of Michigan. This is another billion dollar venture and represents Toyota's third research center, the first two being in Palo Alto, CA, at Stanford, and Cambridge, MA, at MIT. I know this has been reported before, but I like the article. Texas A&M, a school I know pretty much nothing about, is taking a big risk and investing in a huge campus expansion,10 miles from its existing campus, to allow for the type of public-private gown/corporate partnership that Toyota is doing in Michigan. But the Texas plan is a "build it and they will come" venture to be built at a World War II-era site. This will be a $150 million investment. Wise investments as the giant is ahead? Meanwhile Google is planning on making driverless minivans with Chrysler. The signatures, however, are not yet on any documents. When the documents are signed, that deal will fuel driverless media speculation for a while.
Google is reportedly finalizing talks with Fiat Chrysler to manufacture Google driverless cars. This would make available everyday kind of cars as well as luxury models for driverless transport, although none of these are accessible for many people with mobility disabilities. This puts Fiat Chrysler in the driverless game, where they have not yet staked a claim. Update: The Google/Fiat Chrysler deal has been signed. My fantasy in the paragraph above is a bit beyond the first step that the two companies have planned. The agreement signed paves the way for Google to:
develop about 100 self-driving prototypes based on the Chrysler Pacifica hybrid-powered minivan ... The partners’ engineering teams will work together on the project at a facility in southeastern Michigan.
Nvidia doesn't need lines to draw inside of What is being called a technological breakthrough may place Nvidia ahead of everyone in the driverless transportation quest. Nvidia's system essentially teaches itself, in a way that sounds similar to Hotz's Comma.ai system, but is able to navigate beyond highways on unpaved and unmarked roads. It learns by figuring out humans drive. (I can't help a snarky remark: I hope that means someone responsible and not a 16 year old who is taking a first spin after getting his or her license.) This article, by Steven Max Patterson, is well worth reading because it cogently explains the Nvidia technology. Here's a video. The driving seems so ordinary that the video is boring, except that this driverless car is filmed on a range of different types of roads and in all kinds of weather, even snow. With developments like this, the driverless revolution might hit the roads sooner than many expect.
Warning: This is a long post. The most interesting aspect of this type of public hearing is not generally what people say, but who shows up in the first place. The NHTSA hearing in California, held yesterday at Stanford University, attracted many of the same players as the Apr. 8 hearing held in DC., with quite a few additional speakers. Representatives of people with disabilities were out in force and remain amongst the most enthusiastic supporters of driverless vehicles, with the proviso that those vehicles be designed for full accessibility. Other participants included Google's Chris Urmson, David Strickland of the new Self-Driving Coalition for Safer Streets, and a representative of MADD - Mothers Against Drunk Drivers. General themes A general theme for the day and the one point on which there is complete consensus is that the US should adopt a national framework - whether through a uniform set of state laws or via federal laws and regulations - for the operation of driverless vehicles so that manufacturers, safety advocates, and others do not have to operate with a maze of different, and sometimes conflicting, legal provisions. State-by-state air quality standards were raised as one example where that is the case. An aspect of this call for national standards was something heard as well at the Apr. 8 NHTSA hearing, a plea for technology-neutral standards to allow for flexibility in the development of new technology. One speaker, who was from VQuest and does work for the Department of Defense, used the term "platform agnostic." No one seemed to want driverless vehicles to be tied to any particular technology without ease to advance or change. The VQuest speaker made a related point that we do not want to punish failure; we want to create incentives to experiment and thereby learn from failures. Only if we do so will we promote a willingness to share data and innovate. One more point of consensus is the idea of required black boxes, similar to what is on airplanes, to provide data when crashes occur. Some guy also mentioned white boxes, but I did not catch exactly what that was. I admit that I am willing to spend only so much time on this; not willing to go back to the video of the hearing. A last theme, or, perhaps, more of a common refrain, was the recitation of how everyone, including existing manufacturers of currently non-accessible vehicles, is looking forward to a rosy and accessible future for people with disabilities. A less common refrain, though pretty common, was the mention of benefits to the environment. (FYI: If anyone is looking for a brilliant materials science engineer to work on batteries, I have a daughter who fits the bill. She'll be starting a PhD program in the fall. So proud.) Some safety advocates have selective goggles on - for good and bad Some from the so-called consumer safety community continue to drone on about the need for a human driver as though they never received the mountains of news reports over the last one hundred years that drivers have caused over 90 percent of all motor vehicle crashes, the ones that have killed hundreds of thousands and injured millions more - in the US alone, with many millions affected the world over. One of those casualties last year was someone I knew, a fact that scares me each time one of my daughters drives. What the safety advocates make good points about are needs for (1) real federal standards and not just voluntary compacts among the car companies; 2) required transparency of data about operations of vehicles, interfaces, and crashes and near-crashes; and (3) mandatory alerts for car owners and used car owners, in particular, about vehicle recalls and defects. These advocates paint a picture of traditional car companies manipulating a cushy regulatory environment with lax oversight. Neat speaker shows West Coast vibe One biker dude showed up. I admit that I did not catch his name. He was the only speaker who mentioned something important beyond the actual vehicles, which is the people on streets who will not be inside vehicles, autonomous or otherwise. The biker dude expressed concern that self-driving vehicles be designed to decrease worries on the part of, and actual collisions with, bikers and pedestrians. The MADD representative reminded the panel about human drivers who operate vehicles when they are in no condition to drive. MADD, of course, is most concerned about driving under the influence of alcohol and other inebriating substances, and the representative spouted the scary number 80, which is the average number of times a driver drives drunk before an arrest. The advocacy group looks forward to the day when it no longer has to spout such numbers and be concerned about innocent people harmed because a human in a drunken stupor, or just tired after a drink or two (those are my words), gets behind the wheel. NHTSA panel gets star speakers Two rather prominent individuals in the driverless world appeared at the podium to speak at the NHTSA hearing, Chris Urmson of Google and David Strickland, a former NHTSA Administrator, and now counsel and spokesperson for the newly founded Self-Driving Coalition for Safer Streets. (That name, of course, not suggesting that the coalition has no one behind its metaphoric wheel.) A third speaker with some glamour was the representative of Beverly Hills, California (yes, 90210), Dr. Julian Gold, a member of the city council and former mayor. Urmson is about as relaxed a public speaker as I have ever seen. Unlike the other corporate types at the hearing, Urmson seemed frank and conversational. I also like what he said. It's as if he's been listening to me rant on about driverless vehicles. I'd accuse him of eavesdropping at my office and at my kitchen table, but he has no need. He already knows way more about the technology than I do. Urmson pointed out that we already have unsafe roadways with many, many deaths and injuries. There's no reason, he said, to go slowly and incrementally. He succinctly stated that while everyone agrees that the US needs national standards, there is significant disagreement about the speed of implementing the technology and how to develop it. Testing, Urmson reminded the panel, cannot only be performed on testing courses (and fake roads), but must be done on real roads. He talked about the successful experiences of Google's test vehicles and he rightly stated that accidents involving those vehicles have been minor. He also warned the panel that if the US does not act to provide a national regulatory framework that allows for innovation, testing, and deployment, other countries will pass us by.
Coalition talks - Reminding me of the old EF Hutton commercial Strickland announced his new role on behalf of the new Self-Driving Coalition, made similar remarks to others about safety for all and access for people with disabilities, and then progressed to appropriately calling driverless a "safety game changer" and asking for a broad spectrum of stakeholders. Definitely a DC guy. Anyone outside of DC gets the kooties just hearing that word. But Strickland is correct because that is the way Washington, DC, operates best, with coalitions that represent the full range of - ooh, yes I'm using the word - stakeholders. Strickland, as you can tell, spoke in big tent language, and matter of factly told the panel that driverless operations are "years, not decades, away." Even though anyone with a bit of knowledge knows this, the panel needs to hear it from people like Strickland, who have clout and gravitas. He is about as far from a hipster dude as one can be. Suited, serious, and with credentials that precede him, there is no question that the NHTSA panel took his presence there and his words seriously. Confession: I have been in DC long enough that I hardly wince when I hear the "s" word. That's what you get for spending lots of time in an office two blocks from the White House. Town of palm trees and movie star houses wants driverless transit Beverly Hills is actually the City of Beverly Hills, though it is small and feels more like a neighborhood in the LA area. It is very pretty and wealthy. But it recently pushed itself to the forefront of progressive urbanism by passing a resolution in favor of a plan for driverless transit. Dr. Julian Gold, a former mayor and current council member, talked to the NHTSA panel about first and last-mile challenges once transit rail lines are built and in operation nearby. There will be no new parking garages, so the city is realistically considering options and a driverless transit system seems like a good one. The city plans to host a forum soon as it explores how to go about making this dream a reality. (I can see a logo on the vehicle with nice green and pink hues from the Beverly Hills hotel and a white overlay of a tall palm tree. I'm willing to design it.) Gold also envisions driverless for seniors and people with disabilities who live in his city. He reminded the panel that parking spaces cost about $40,000 a pop. They are not free. Gold was practical in raising issues such as whether children will be allowed to ride with adults (they do on subways and there's no driver or other official in every car), and what roadway and vehicle specifications, if any, will need to be in place. Gold brought the local government perspective, which is an important one in transportation. Local and state governments are responsible for transit and for road maintenance. He informed the panel that these questions need to be answered ASAP. And his city wants the full monty of level five, fully autonomous vehicles. The federal-state division of responsibilities may need to be revisited wholesale instead of globbing on a new layer of regulation onto the old. The Nvidia spokesperson suggested as much in his comments. Truth be told, however, globbing is a traditional DC strategy in a town where consensus is the name of the game. Award for most effective speaker The representative of Self-Driving MN (Minnesota) at yesterday's NHTSA hearing spoke from the heart as he described the hi-jinks of family members and a mix of family and public transportation that provide transportation to work for his brother who is disabled. He spoke eloquently when he wondered aloud how someone with a disability who does not have family close by is able to get to work and to otherwise take care of oneself. The intimate scene that this gentleman allowed everyone watching the hearing to share painted a before and after picture that demonstrated the freedom and equality that autonomous vehicles will represent for many people with disabilities. Another representative of the disability community reminded panel members that blindness and visual impairments are not the only class of disability that will require accessibility to driverless vehicles, a point emphasized in more ways at the Apr. 8 hearing in DC. However, that same speaker, whose name I did not jot down, aptly uttered the mantra of the disability community - nothing about us without us - when calling for universal design to be incorporated in and interfacing with these new vehicles at the outset. Award for speaker who probably knows a thing or two about microbreweries The Zoox speaker was by far the coolest dude in the room, with a sense of humor and a let's-figure-this-out-over-a-beer vibe. He also had on a nice hat and told the story of how his company got started just a stone's throw away from where the NHTSA hearing was taking place. It made me smile. He got to the heart of the matter at hand for NHTSA, which is how to have technology experts and government regulators, who approach their lines of work very differently, to come together for the common good. And he used the word traumatized to describe the experiences of families around the globe who have lost loved ones. A big thank you to the tweeter, Mark Harris (@meharris, published in the Economist, among others), who pointed out that I was wrong with my spelling of Zoox. I was thinking in terms of rhymes with cukes. Usual suspects Traditional car companies showed up as well, including Ford, Honda, Toyota, and GM. No surprises there. No news - with one exception. Toyota is pushing for a transitional period, which, when translated, means, in my opinion, that the company is not ready for prime time on driverless and it does not want anyone else getting out of the gate on this any time soon. Likewise for the speaker from Subaru. (Okay, can we really trust a car company with a dweeb advertising campaign that it's cars are made with love? I thought they were made with robots and non-union labor.) Best corporate speaker award This prize definitely goes to the Volvo speaker, who right off the bat mentioned that she has three children and who managed to humanize the Volvo corporate goals of zero deaths by 2020 and its acceptance of full responsibility - meaning liability - for any crashes involving its driverless vehicles, which, by the way, will be on London roads soon, and, of course, Swedish ones. (Trust me, London can compete with New York for crazy driving, though I believe that New Yorkers are less about following rules and definitely more about shouting out the window in an endless loud back and forth with pedestrians, except above Mitdtown, where there is less and less traffic.) If you would like to watch the Apr. 8 hearing, here is the video. I suspect the video of the Apr. 27 hearing will be available soon.
In case anyone wants more of a say, NHTSA will be accepting public comments until May 9, the day after Mother's Day.
Singapore + nuTonomy = Pod-mania OMG, the twittersphere and conventional journalism - here, via Forbes - are full of Singapore's plan to introduce taxi-pods (taxibots?). These will be accessible by apps. How else, right? (Not quicker, by the way, than catching a cab on CPW in Manhattan.) This could beat Uber at its own game and get a leg up on shared-use plans by Google, GM, Ford, and others.
Google stays out West Google is bringing its cute driverless vehicles to Phoenix, Arizona. Who wants to be there? No offense, but it' not a pedestrian-friendly place. Maybe it's just my East Coast sensibilities. Google says there's good air-born dust, as in dust storms. I was there for one once and it was creepy. Thank goodness someone else was driving. North Dakota to have driverless highway? Every hardly-any-people state has two senators
One weird aspect of the US Constitution is that every state, no matter how small and no matter how few people, gets two senators. Hence a state like Wyoming, which has a population the size of a Brooklyn neighborhood, gets somewhat outsize representation. In a way, this is a good thing. Some of those senators are able to push through interesting ideas and not get swamped by the more populous states. North Dakota (pop. a bit under 750,000) Sen. John Thune is backing an idea to allow a particular highway that runs north-to-south and goes through seven states to be the first to allow driverless hauling of freight. Lots of driverless trucks. There's an association formed to push the idea toward fruition, the Central North Trade Corridor Association. This would mean seven states permitting driverless travel, whether through laws, regulations, or pure neglect. It could just mean interpreting existing provisions to not require a human driver. They're all laissez-faire/no-government-intervention out West, (super broad generalization there) so maybe that could happen. Truck trip success in Europe Speaking of a trucking company fantasy, a convoy of connected trucks - lorries, to those of you in the UK - has completed a trip across Europe. I think the first truck needs a driver, but the followers do not, though, of course, there were drivers present during this test. I'm not sure what happens to the CV trucks when obnoxious, speeding drivers of cars weave into the middle of the convoy. That will be the New Jersey Turnpike test. Volvo going to China First, Volvo is now owned by a Chinese company. Second, Chinese companies are pushing for their government to establish driverless-friendly rules. Third, China presents good challenges for driverless vehicles, as in crazy congestion and drivers. Volvo plans to essentially give 100 ordinary people 100 driverless vehicles to test on urban Chinese roads. I think the drivers must be present and they will be able to take over the wheel. This pilot will not happen until 2017. NHTSA and technical details that make driverless possible Tomorrow, the US National Highway Traffic Safety Administration (NHTSA) will be hosting a public meeting at the USDOT (Department of Transportation) building. One of those details is the technology that allows driverless vehicles to "see" its surroundings and "communicate" - as in connected vehicles or CV - with others. It will be interesting because DOT announced it will come out with regulations within the next few months. Okay, I remember that, but I don't have a link right now. That is warp speed for a government agency.
Two pretty quiet but big developments as the world edges ever closer to the Age of Driverless. 1. The first privately-owned Google driverless vehicle has been purchased by the Scottish "ideas agency" company Equator. This is not a test. The driverless car will be ferrying actual people, staff and clients (not Google-paid "drivers") that is, from Equator offices in Finnieston (not sure if that's a different town or a neighborhood) to the Glasgow offices of its sister design agency 999. 2. The Capital Beltway, the road that practically defines the inner and outer people and areas of the Washington, DC, region, hosted a driverless test of a large vehicle, in regular traffic. No video of this driverless trip has been released. The company doing the test was AiNET, based in Beltsville, MD, right by the Beltway. The now self-driving vehicle was a reclaimed 1991 Amertek Aircrash Rescue and Fire Fighting vehicle. Yes, a fire truck that drives itself. Now all of the firefighters can wave to passersby. Cute vehice, no?
We're talking real traffic And there is traffic at every time of a weekday on this notorious road. This is a road with lots of road splits and interchanges with other roads, and it is a road that is pretty round, sometimes with a strange shape, sometimes near a building that looks mysteriously similar to the castle in the Wizard of Oz. For trivia's sake, I must mention that for many years, right by that castle (which is actually a Mormon temple, I believe), these words appeared on a Beltway overpass: Surrender Dorothy. This test was a 50-mile trip in Maryland on the Beltway and then through neighborhoods and into Washington, DC, into such tony, fashionable areas as Georgetown, Capital Hill, and Shaw, which is gentrifying from hipster to fashionable.
Alaska: Google sends emissary Ron Barnes, its state legislative representative, to flirt with Alaskan officials. It's mutual flirting at this point, with no commitments, between Google and Alaska to decide whether to test Google cars in the frozen state. Alaska provides lots of snow and ice and quiet roadways. US Department of Transportation: Not exactly driverless, but driverless-related news, in Google's participation in the DOT Smart Cities' challenge. Google will be collecting and aggregating vehicular travel data in cities, which data will then be plowed into improving driverless vehicle capability. Here's an explanation of the arrangement.
The project, called Flow, is part of a joint venture between the DOT and Sidewalk Labs, a company run under Google-turned-Alphabet, to aggregate anonymized data to better understand traffic patterns and congestion areas in cities.
Over in Michigan, another dance tune The state of Michigan and GM are getting more serious about testing and developing self-driving vehicles. In a addition to MCity, the state will have a new testing facility, Willow Run, to be headed by a University of Michigan official and to be fitted up with some state money. The official headed up the program that oversaw MCity. More information here. Dancing alone, Ford announces the creation of its subsidiary, Smart Mobility. Ho hum news, really.
