Audi

This self-driving AI faced off against a champion racer (kind of)

Posted by | artificial intelligence, Audi, automotive, Gadgets, hardware, robotics, science, self-driving cars, stanford, Stanford University, Transportation | No Comments

Developments in the self-driving car world can sometimes be a bit dry: a million miles without an accident, a 10 percent increase in pedestrian detection range, and so on. But this research has both an interesting idea behind it and a surprisingly hands-on method of testing: pitting the vehicle against a real racing driver on a course.

To set expectations here, this isn’t some stunt, it’s actually warranted given the nature of the research, and it’s not like they were trading positions, jockeying for entry lines, and generally rubbing bumpers. They went separately, and the researcher, whom I contacted, politely declined to provide the actual lap times. This is science, people. Please!

The question which Nathan Spielberg and his colleagues at Stanford were interested in answering has to do with an autonomous vehicle operating under extreme conditions. The simple fact is that a huge proportion of the miles driven by these systems are at normal speeds, in good conditions. And most obstacle encounters are similarly ordinary.

If the worst should happen and a car needs to exceed these ordinary bounds of handling — specifically friction limits — can it be trusted to do so? And how would you build an AI agent that can do so?

The researchers’ paper, published today in the journal Science Robotics, begins with the assumption that a physics-based model just isn’t adequate for the job. These are computer models that simulate the car’s motion in terms of weight, speed, road surface, and other conditions. But they are necessarily simplified and their assumptions are of the type to produce increasingly inaccurate results as values exceed ordinary limits.

Imagine if such a simulator simplified each wheel to a point or line when during a slide it is highly important which side of the tire is experiencing the most friction. Such detailed simulations are beyond the ability of current hardware to do quickly or accurately enough. But the results of such simulations can be summarized into an input and output, and that data can be fed into a neural network — one that turns out to be remarkably good at taking turns.

The simulation provides the basics of how a car of this make and weight should move when it is going at speed X and needs to turn at angle Y — obviously it’s more complicated than that, but you get the idea. It’s fairly basic. The model then consults its training, but is also informed by the real-world results, which may perhaps differ from theory.

So the car goes into a turn knowing that, theoretically, it should have to move the wheel this much to the left, then this much more at this point, and so on. But the sensors in the car report that despite this, the car is drifting a bit off the intended line — and this input is taken into account, causing the agent to turn the wheel a bit more, or less, or whatever the case may be.

And where does the racing driver come into it, you ask? Well, the researchers needed to compare the car’s performance with a human driver who knows from experience how to control a car at its friction limits, and that’s pretty much the definition of a racer. If your tires aren’t hot, you’re probably going too slow.

The team had the racer (a “champion amateur race car driver,” as they put it) drive around the Thunderhill Raceway Park in California, then sent Shelley — their modified, self-driving 2009 Audi TTS — around as well, ten times each. And it wasn’t a relaxing Sunday ramble. As the paper reads:

Both the automated vehicle and human participant attempted to complete the course in the minimum amount of time. This consisted of driving at accelerations nearing 0.95g while tracking a minimum time racing trajectory at the the physical limits of tire adhesion. At this combined level of longitudinal and lateral acceleration, the vehicle was able to approach speeds of 95 miles per hour (mph) on portions of the track.

Even under these extreme driving conditions, the controller was able to consistently track the racing line with the mean path tracking error below 40 cm everywhere on the track.

In other words, while pulling a G and hitting 95, the self-driving Audi was never more than a foot and a half off its ideal racing line. The human driver had much wider variation, but this is by no means considered an error — they were changing the line for their own reasons.

“We focused on a segment of the track with a variety of turns that provided the comparison we needed and allowed us to gather more data sets,” wrote Spielberg in an email to TechCrunch. “We have done full lap comparisons and the same trends hold. Shelley has an advantage of consistency while the human drivers have the advantage of changing their line as the car changes, something we are currently implementing.”

Shelley showed far lower variation in its times than the racer, but the racer also posted considerably lower times on several laps. The averages for the segments evaluated were about comparable, with a slight edge going to the human.

This is pretty impressive considering the simplicity of the self-driving model. It had very little real-world knowledge going into its systems, mostly the results of a simulation giving it an approximate idea of how it ought to be handling moment by moment. And its feedback was very limited — it didn’t have access to all the advanced telemetry that self-driving systems often use to flesh out the scene.

The conclusion is that this type of approach, with a relatively simple model controlling the car beyond ordinary handling conditions, is promising. It would need to be tweaked for each surface and setup — obviously a rear-wheel-drive car on a dirt road would be different than front-wheel on tarmac. How best to create and test such models is a matter for future investigation, though the team seemed confident it was a mere engineering challenge.

The experiment was undertaken in order to pursue the still-distant goal of self-driving cars being superior to humans on all driving tasks. The results from these early tests are promising, but there’s still a long way to go before an AV can take on a pro head-to-head. But I look forward to the occasion.

Powered by WPeMatico

Transportation Weekly: Polestar CEO speaks, Tesla terminology, and a tribute

Posted by | alex roy, Android, Aptiv, Audi, Automation, automotive, BMW, Canada, car, car sharing, Carmera, cars, China, e-bikes, Environmental Protection Agency, Ford, Google, honda, Joshua Schachter, Kia, Kirsten Korosec, Las Vegas, Lyft, mobility services, Netflix, New York, Peugeot, pininfarina, Polestar, rakuten, self-driving car, sidewalk labs, simulation, TechCrunch, Tesla Model S, tokyo, toronto, Toyota, toyota research institute, Transportation, Transportation Weekly, volkswagen, volvo, waymo, Zipcar, Zum | No Comments

Welcome back to Transportation Weekly; I’m your host Kirsten Korosec, senior transportation reporter at TechCrunch . This is the fourth edition of our newsletter, a weekly jaunt into the wonderful world of transportation and how we (and our packages) move.

This week we chat with Polestar CEO Thomas Ingenlath, dig into Lyft’s S-1, take note of an emerging trend in AV development, and check out an experiment with paving. Oh, and how could we forget Tesla.

Never heard of TechCrunch’s Transportation Weekly? Catch up here, here and here. As I’ve written before, consider this a soft launch. Follow me on Twitter @kirstenkorosec to ensure you see it each week. (An email subscription is coming). 


ONM …

There are OEMs in the automotive world. And here, (wait for it) there are ONMs — original news manufacturers. (Cymbal clash!) This is where investigative reporting, enterprise pieces and analysis on transportation lives.

This week, we’re featuring excerpts taken from a one-on-one interview with Polestar CEO Thomas Ingenlath.

On February 27, Volvo’s standalone electric performance brand Polestar introduced its first all-electric vehicle, a five-door fastback called the Polestar 2. The EV, which has a 78 kWh battery pack and can travel 275 miles (estimated EPA guidance) on a single charge, will be manufactured at a new factory in Chengdu, China. Other notable specs: The infotainment system will be powered by Android OS, Polestar is offering subscriptions to the vehicle, and production starts in 2020.

yellow-jacket-polestar

Here is what Ingenlath had to say to me about …

EV charging infrastructure

To be very unpolitical, I think it would be totally stupid if we were to aim to develop electric charging infrastructure on our own or for our brand specifically. If you join the electric market today, of course, you would see partnerships; that’s sensible thing to do. Car companies together are making a big effort in getting out a network of necessary charging stations along the highway. 

That’s what we’re doing; we’re teaming up and have the contracts being designed and soon signed.

On the company’s approach to automation 

The terminology is important for us. We very clearly put that into a different picture, we’re not talking about, and we clearly do not ever want to label it, anautopilot.” The focus of this system is a very safe distance control, which brakes for you and accelerates for you, and of course, the lane keeping. This is not about developing an autopilot system, it is about giving your safety. And that’s where we don’t want to provoke people thinking that they have full rollout autopilot system there. But it is a system that helps you being safe and protected on the road.

I also reached out to Transportation Weekly readers and asked what they wanted to know and then sent some of those questions to Ingenlath.

TW Reader: How did it feel taking one of your personal styling elements – the C shaped rear lamps – from your previous brand over to Polestar?
Ingenlath: It’s an evolutionary process. Polestar naturally builds on its “mothers” DNA and as a new branch develops its own personality. Thor’s hammer, the rear light signature -—with each new model launch (Volvo and Polestar) those elements diverge into a brand specific species.
TW Reader: How much do you still get to do what you love, which is design?
Ingenlath: Being creative is still my main job, now applied on a broader scope — trying to lead a company with a creative and  brand building mindset. Still, I love the Fridays when I meet up with Robin and Max to review the models, sketches and new data. We really enjoy driving the design of both brands to new adventures.

Dig In

Tesla is finally going to offer customers a $35,000 Model 3. How the automaker is able to sell this electric vehicle at the long-awaited $35,000 price point is a big piece of that story — and one that some overlooked. In short, the company is blowing up its sales model and moving to an online only strategy. Tesla stores will close or be converted to “information centers” and retail employees will be laid off.

But this is not what we’re going to talk about today. Tesla has also brought back its so-called “full self-driving” feature, which was removed as an option on its website last year. Now it’s back. Owners can opt for Autopilot, which has automatic steering on highways and traffic-aware cruise control, or FSD.

FSD capability includes several features such as Navigate on Autopilot that is supposed to guide a car from a highway on-ramp to off-ramp, including navigating interchanges and making lane changes. FSD also includes Advanced Summon, Auto Lane Change, and Autopark. Later this year, the system will recognize and respond to traffic lights in more complex urban environments, Tesla says.

All of these features require the driver to be engaged (or ready to take over), yet it’s called “full self-driving.” Now Tesla has two controversially named automation features. (The other is Autopilot). As Andrew Hawkins at The Verge noted in his coverage, “experts are beginning to realize that the way we discuss, and how companies market, autonomy is significant.”

Which begs the obvious question, and one that I asked Musk during a conference call on Thursday. “Isn’t it a problem that you’re calling this full self-driving capability when you’re still going to require the driver to take control or be paying attention?” (I also wanted to ask a followup on his response, but the moderator moved onto the next reporter).

His response:

“We are very clear when you buy the car what is meant by full self driving. It means it’s feature complete, but feature complete requiring supervision.

As we get more — we really need billions of miles, if not maybe 10 billion sort of miles or kilometers on that order collectively from the fleet — then in our opinion probably at that point supervision is not required, but that will still be up to regulators to agree.

So we’re just very clear.  There’s really three steps: there’s being feature complete of full self driving that requires supervision, feature complete but not requiring supervision, and feature complete not requiring supervision and regulators agree.

In other Tesla news, the National Transportation Safety Board is investigating a crash, that at first glance seems to be similar to the fatal crash that killed Tesla owner Joshua Brown.

In cooperation with the Palm Beach sheriff’s office, the NTSB is sending a team of three to conduct a safety investigation of the commercial motor vehicle and Tesla crash in Delray Beach, FL.

— NTSB_Newsroom (@NTSB_Newsroom) March 2, 2019


A little bird …

We hear a lot. But we’re not selfish. Let’s share.

blinky-cat-bird

It’s no secret that Pittsburgh is one of the hubs of autonomous vehicle development in the world. But what’s not so widely known — except for a group of government and company insiders — is that Mayor William Peduto is on the verge of issuing an executive order that will give more visibility into testing there. 

The city’s department of mobility and infrastructure is the central coordinator of this new executive order that aims to help guide testing and policy development there. The department is going to develop guidelines for AV testing, we’re told. And it appears that information on testing will be released to the public at least once a year.

Got a tip or overheard something in the world of transportation? Email me or send a direct message to @kirstenkorosec.


Deal of the week

Daimler and BMW are supposed to be competitors. And they are, except with mapping (both part of the HERE consortium), mobility services (car sharing, ride-sharing), and now the development of highly automated driving systems. The deal is notable because it illustrates a larger trend that has emerged as the AV industry hunkers down into the “trough of disillusionment.” And that’s consolidation. If 2016, was the year of splashy acquisitions, then 2019 is shaping up to be chockfull of alliances and failures (of some startups).

Also interesting to note, and one that will make some AV safety experts cringe, both companies are working on Level 3 driving automation, a designation by the SAE that means conditional driving automation in which multiple high levels of automation are available in certain conditions, but a human driver must be ready to take over. This level of automation is the most controversial because of the so-called “hand off” problem in which a human driver is expected to take control of the wheel in time.

Speaking of partnerships, another deal that got our attention this week involved New York-based mapping and data analytics startup Carmera and Toyota Research Institute-Advanced Development. TRI-AD is an autonomous drive unit started by Toyota with Denso and Aisin. TRI-AD’s mission is to take the research being done over at the Toyota Research Institute and turn its into a product.

The two companies are going to test a concept that will use cameras in Toyota test vehicles to collect data from downtown Tokyo and use it to create high definition maps for urban and surface roads.

TRI-AD considers this the first step towards its open software platform concept known as Automated Mapping Platform that will be used to support the scalability of highly automated driving, by combining data gathered from vehicles of participating companies to generate HD maps. AMP is new and has possible widespread implications at Toyota. And TRI-AD is full of A-listers, including CEO James Kuffner, who came from the Google self-driving project and Nikos Michalakis, who built Netflix’s cloud platform, and Mandali Khalesi, who was at HERE.

Read more on Khalesi and the Toyota’s open source ambitions here.

Other deals:


Snapshot

Snapshot this week is a bit untraditional. It’s literally a snapshot of myself and my grandmother, months before her 100th birthday. Her memorial service was held Saturday. She died at 101. She loved cars and fast ones, but not so much driving them. And every time I got a new press car, we’d hit the road and she’d encourage me to take the turns a bit faster.

She also loved road trips and in the 1920s, her father would drive the family on the mostly dirt roads from New Jersey to Vermont and even Canada. In her teens, she loved riding in the rumble seat, a feature found in a few vehicles at the time including the Ford Model A.

She was young at heart, until the very end. Next week, we’ll focus on the youngest drivers and one automotive startup that is targeting that demographic.


Tiny but mighty micromobility

Lyft’s S-1 lays out the risks associated with its micromobility business and its intent to continue relying on third parties to manufacture its bikes and scooters. Here’s a key nugget about adoption:

“While some major cities have widely adopted bike and scooter sharing, there can be no assurance that new markets we enter will accept, or existing markets will continue to accept, bike and scooter sharing, and even if they do, that we will be able to execute on our business strategy or that our related offerings will be successful in such markets. Even if we are able to successfully develop and implement our network of shared bikes and scooters, there may be heightened public skepticism of this nascent service offering.”

And another about seasonality:

“Our limited operating history makes it difficult for us to assess the exact nature or extent of the effects of seasonality on our network of shared bikes and scooters, however, we expect the demand for our bike and scooter rentals to decline over the winter season and increase during more temperate and dry seasons.”

Lyft, which bought bike-share company Motivate back in July, also released some data about its electric pedal-assist bikes this week, showing that the pedal assist bikes are, unsurprisingly, more popular than the traditional bikes. They also traveled longer distances and improved winter ridership numbers. Now, Lyft is gearing up to deploy 4,000 additional electric bikes to the Citi Bike system in New York City.

One more thing …

Google Maps has added a feature that lets users see Lime scooters, pedal bikes and e-bikes right from the transit tab in over 80 new cities around the world. Users can click the tab to find out if Lime vehicle is available, how long it’ll take to walk to the vehicle, an estimate of how much their ride could cost, along with total journey time and ETA.


Notable reads

If take the time to read anything this week (besides this newsletter), spend some time with Lyft’s S-1. The ride-hailing company’s prospectus mentions autonomous 109 times. In short, yeah, it’s something the company’s executives are thinking about and investing in.

Lyft says it has a two-pronged strategy to bring autonomous vehicles to market. The company encouraging developers of autonomous vehicle technology to use its open platform to get access to its network and enable their vehicles to fulfill rides on the Lyft platform. And Lyft is trying to build its own autonomous vehicle system at its confusingly named “Level 5 Engineering Center.”

  • The company’s primary investors are Rakuten with a 13 percent stake, GM with 7.8 percent, Fidelity with 7.7 percent, Andreessen Horowitz with 6.3 percent and Alphabet with 5.3 percent. GM and Alphabet have business units, GM Cruise and Waymo respectively, that are also developing AV technology.
  • Through Lyft’s partnership with AV systems developer and supplier Aptiv, people in Las Vegas have taken more than 35,000 rides in Aptiv autonomous vehicles with a safety driver since January 2018.
  • One of the “risks” the company lists is “a failure to detect a defect in our autonomous vehicles or our bikes or scooters”

Other quotable notables:

Check out the Pedestrian Traffic Fatalities by State report, a newly released report from Volvo Car USA and The Harris Poll called  The State of Electric Vehicles in America.


Testing and deployments

Again, deployments doesn’t always mean the latest autonomous vehicle pilot.

On Saturday, Sidewalk Labs hosted its Open Sidewalk event in Toronto. This is part of Sidewalk Toronto, a joint effort by Waterfront Toronto and Alphabet’s Sidewalk Labs to create a “mixed-use, complete community” on Toronto’s Eastern Waterfront

The idea of this event was to share ideas and prototypes for making outdoor public space the “social default year-round.” One such prototype “hexagonal paving” got our attention because of its use case for traffic control and pedestrian and bicyclist safety. (Pictured below)

These individual precast concrete slabs are movable and permeable, can light up and give off heat. The idea is that these hexagonal-shaped slabs and be used to clear snow and ice in trouble spots and light up to warn drivers and pedestrians of changes to the street use or to illuminate an area for public uses or even designate bike lanes and hazard zones. And because they’re permeable they can be used to absorb stormwater or melted snow and guide it to underground stormwater management systems.

Sidewalk Labs tell me that the pavers have “plug and play” holes, which allow things like bike racks, bollards, and sign posts to be inserted. Sidewalk Labs initially built these with wood, and the new prototype is the next iteration, featuring modules built from concrete.


On our radar

There is a lot of transportation-related activity this month.

The Geneva Motor Show: Press days are March 5 and March 6. Expect concept, prototype and production electric vehicles from Audi, Honda, Kia, Peugeot, Pininfarina, Polestar, Spanish car company Hispano Suiza, and Volkswagen.

SXSW in Austin: TechCrunch will be at SXSW this coming week. Here’s where I’ll be.

  • 2 p.m. to 6:30 p.m. March 9 at the Empire Garage for the Smart Mobility Summit, an annual event put on by Wards Intelligence and C3 Group. The Autonocast, the podcast I co-host with Alex Roy and Ed Niedermeyer, will also be on hand.
  • 9:30 a.m. to 10:30 a.m. March 12 at the JW Marriott. The Autonocast and founding general partner of Trucks VC, Reilly Brennan will hold a SXSW podcast panel on automated vehicle terminology and other stuff.
  • 3:30 p.m over at the Hilton Austin Downtown, I’ll be moderating a panel Re-inventing the Wheel: Own, Rent, Share, Subscribe. Sherrill Kaplan with Zipcar, Amber Quist, with Silvercar and Russell Lemmer with Dealerware will join me.
  • TechCrunch is also hosting a SXSW party from 1 pm to 4 pm Sunday, March 10, 615 Red River St., that will feature musical guest Elderbrook. RSVP here

Self Racing Cars

Finally, I’ve been in contact with Joshua Schachter who puts on the annual Self Racing Car event, which will be held March 23 and March 24 at Thunderhill Raceway near Willows, California.

There is still room for participants to test or demo their autonomous vehicles, drive train innovation, simulation, software, teleoperation, and sensors. Hobbyists are welcome. Sign up to participate or drop them a line at contact@selfracingcars.com.

Thanks for reading. There might be content you like or something you hate. Feel free to reach out to me at kirsten.korosec@techcrunch.com to share those thoughts, opinions or tips. 

Nos vemos la próxima vez.

Powered by WPeMatico

Holoride’s in-car VR solution is the best thing at CES 2019

Posted by | Audi, automotive, CES 2019, Gadgets, holoride, Virtual reality | No Comments

After days of demos and announcements and miles of walking, I’m confident in declaring Holoride the best thing at this year’s CES. The designation of “The best thing at CES 2019” is my badging. This isn’t an official award handed out by a governing body. This is just me saying Holoride is the best thing I’ve seen at the show.

This year’s CES is fine, I guess. The main theme is connecting services around the smart home. There’s a huge range of devices that now support services from Amazon, Google and Apple. CES 2019 also featured the launch of new silicon chipsets and self-driving platforms. But the thing that impressed me the most is from Holoride, a startup from Audi that wants to put VR in cars to entertain and reduce motion sickness.

Iron Man needs help, Rocket told me. And like that I was thrust into a space battle against Thanos’ bad guys. There was an Oculus on my head and my body was dipping and diving, shooting through space, while I was waving my hands around, blasting the enemy. It was straight out of Disney World (partly because Disney helped with the content). Except I was in Vegas, in the back of an Audi SUV hitting speeds of 90 mph on a track.

After two laps around the track, I walked away fine. I didn’t feel sick at all, even though I’m the sort of person who can’t look at their phone in a car.

Matching the VR content to the vehicle’s movements is key to the Holoride experience. In short, when the car moves, the content moves in the same way. This reduces motion sickness, and, from my demo, I can confirm it works — at least on me.

The technology comes from a small startup recently spun out of Audi in a play to put VR in every car. The founders have been working on the technology behind the in-car VR system for several years. The automaker holds a minority interest through subsidiary Audi Electronics Venture, which helped develop the technology. Audi will license the technology to Holoride and the startup will use an open platform to allow any automaker as well as content developers to create whatever reality formats they desire.

I’ve experienced countless VR experiences, and this was one of the best demos I’ve had. The use case is compelling too. Not only does it provide entertainment, but it also solves motion sickness. It’s easy to imagine this in an ad-supported format in the back of an Uber or while on a long-distance bus. It could work in planes too. It could improve long car rides with the kids.

Holoride is a longshot and there are countless questions around the content, consumer outreach and compatibility. In order for it to take off, the company needs to build an ecosystem complete with developers, auto makers and consumers. Building amazing experiences is one thing; selling amazing experiences is even harder.

CES 2019 coverage - TechCrunch

Powered by WPeMatico

Volvo and Audi turn to Android for next-gen connected cars

Posted by | Android, Audi, automotive, Google, TC, Transportation, volvo | No Comments

 Volvo and Audi are looking to do more with Android in its forthcoming connected vehicles, the companies announced on Monday ahead of this week’s Google I/O developer conference. The companies will be showing off the fruits of this deeper integration at the event in San Jose, which goes beyond entertainment and navigation to help drivers and passengers control AC, windows, sunroof and… Read More

Powered by WPeMatico

To the Moon! Lunar XPRIZE team looks to send Wikipedia into space aboard homemade rover

Posted by | Audi, Education, Europe, Gadgets, Google, google lunar x prize, science, Social, Space, TC, Wikimedia, Wikipedia, XPRIZE | No Comments

PTS lunar rover Quick: you can send 20 gigabytes of data to the moon. What’s it going to be? Time’s up!
Wikipedia? What a coincidence — that’s what the Part-Time Scientists, a team working on a homegrown lunar rover for Google’s Lunar XPRIZE, decided too! And they’re really planning to bring Wikipedia to the Moon. Read More

Powered by WPeMatico

Audi Leads $28M Investment In Rental Startup Silvercar

Posted by | Audi, ces2016, Mobile, Silvercar, Startups, TC | No Comments

silvercar austin Silvercar, a startup rethinking the auto rental experience in airports, already seems pretty tied to Audi — after all, every vehicle that Silvercar rents out is a silver Audi. Now the companies are deepening that relationship with a $28 million Series C investment.
Audi led the round, with the company’s North American president Scott Keogh joining Silvercar’s board of directors. Read More

Powered by WPeMatico