Menu

Timesdelhi.com

June 16, 2019
Category archive

Tesla

Consumer Reports knocks Tesla’s Navigate on Autopilot feature

in Automation/Automotive/Autopilot/consumer reports/Delhi/India/Politics/Tesla/tesla model 3/Tesla Model S/Transportation by

Consumer Reports is calling the automatic lane-change feature on Tesla’s Navigate on Autopilot “far less competent” than a human driver and cautioned it could pose safety risks.

The consumer advocacy organization posted its review Wednesday on the newest version of Tesla’s advanced driver assistance system.

Navigate on Autopilot is an active guidance system that is supposed to navigate a car from a highway on-ramp to off-ramp, including interchanges and making lane changes. Once drivers enter a destination into the navigation system, they can enable “Navigate on Autopilot” for that trip.

Tesla pushed out a software update last month to allow for automatic lane changes. Drivers have to enable this feature, which gives the car permission to make its own lane changes. If not enabled, the system asks the driver to confirm the lane change before moving over. Automatic lane changes can be canceled at any time.

The system has been touted as a way to make driving less stressful and improve safety. In practice, the system had startling behavior, Jake Fisher, senior director of auto testing at Consumer Reports told TechCrunch.

“It doesn’t take very long behind the wheel with this feature on to realize it’s not quite ready for prime time,” Fisher said. CR said one of the more troubling concerns were failures of Tesla’s three rearward-facing cameras to detect fast-approaching objects from the rear better than the average driver.

The CR reviewers found Navigate on Autopilot lagged behind human driving skills and engaged in problematic behavior such as cutting off cars and passing on the right. CR drivers often had to take over to prevent the system from making poor decisions.

As a result, the system increases stress and doesn’t improve safety, Fisher said, before asking “So what is the point of this feature?”

The automatic lane change reviewed by Consumer Reports is not the default setting for Autopilot, Tesla notes. It’s an option that requires drivers to remove the default setting. Tesla also argues that drivers using Navigate on Autopilot properly have successfully driven millions of miles and safely made millions of automated lane changes.

While Fisher acknowledged the default setting, he contends that isn’t the issue. He notes the Tesla has many warnings that the driver must be alert and ready to take over at any time.

“Our concern is that if you’re not alert (or ready to take over) you could be put into a tricky situation,” he said.

The bigger concern for all systems like these is the driver will put too much trust into it, Fisher said. The automatic lane-change feature might not be good enough for drivers to let down their guard yet. If Tesla improves this system, even a little bit, the risk of complacency and too much trust rises.

And that’s problematic because drivers still must be ready to take over. “Just watching automation is a harder human task than driving the car,” he said.

CR asserts that an effective driver monitoring system would mitigate this risk. DMS is typically a camera combined with software designed to track a driver’s attention and pick up on cognitive issues that could cause an accident such as drowsiness.

DMS are found in certain BMW models with an ADAS system called DriverAssist Plus, the new 2020 Subaru Outback and Cadillac’s equipped with its Super Cruise system.

This isn’t the first time CR has raised concerns about Autopilot. Last week, the consumer advocacy organization called on Tesla to restrict the use of Autopilot and install a more effective system to verify driver engagement in response to a preliminary report by National Transportation Safety Board on the fatal March 2019 crash of a Tesla Model 3 with a semi-trailer in Delray Beach, Fla.

Last year, CR gave GM’s Super Cruise the top spot in its first-ever ranking of partially automated driving systems because it is the best at striking a balance between technical capabilities and ensuring drivers are paying attention and operating the vehicle safely. Tesla followed in the ranking not because it was less capable, but because of its approach to safety, Fisher noted.

CR evaluated four systems: Super Cruise on the Cadillac CT6, Autopilot on Tesla Model S, X and 3 models, ProPilot Assist on Infiniti QX50 and Nissan Leaf, and Pilot Assist on Volvo XC40 and XC60 vehicles. The organization said it picked these systems because they’re considered the most capable and well-known in the industry.

Part fund, part accelerator, Contrary Capital invests in student entrepreneurs

in austin/california/Columbia University/Contrary Capital/Dan Macklin/Delhi/Dorm Room Fund/Economy/Emmett Shear/engineer/Entrepreneurship/Facebook/Finance/First Round Capital/General Catalyst/Graduate Fund/harvard/India/martin eberhard/MIT/mulesoft/pennsylvania/Politics/Private equity/productivity software/Prototype Capital/Reddit/San Francisco/SoFi/stanford/Stanford University/Startup company/Startups/steve huffman/Tesla/texas/Twitch/United States/university of california san diego/University of Pennsylvania/university of texas/Venture Capital/Y Combinator by

First Round Capital has both the Dorm Room Fund and the Graduate Fund. General Catalyst has Rough Draft Ventures. And Prototype Capital and a few other micro-funds focus on investing in student founders, but overall, there’s a shortage of capital set aside for entrepreneurs still making their way through school.

Contrary Capital, a soon-to-be San Francisco-based operation led by Eric Tarczynski, is raising $35 million to invest between $50,000 and $200,000 in students and recent college dropouts. The firm, which operates a summer accelerator program for its portfolio companies, closed on $2.2 million for its debut, proof-of-concept fund in 2018.

“We really care about the founders building a great company who don’t have the proverbial rich uncle,” Tarczynski, a former founder and startup employee, told TechCrunch. “We thought, ‘What if there was a fund that could democratize access to both world-class capital and mentorship, and really increase the probability of success for bright university-based founders wherever they are?’ “

Contrary launched in 2016 with backing from Tesla co-founder Martin Eberhard, Reddit co-founder Steve Huffman, SoFi co-founder Dan Macklin, Twitch co-founder Emmett Shear, founding Facebook engineer Jeff Rothschild and MuleSoft founder Ross Mason. The firm has more than 100 “venture partners,” or entrepreneurial students at dozens of college campuses that help fill Contrary’s pipeline of deals.

Contrary Capital celebrating its Demo Day event last year

Last year, Contrary kicked off its summer accelerator, tapping 10 university-started companies to complete a Y Combinator -style program that culminates with a small, GP-only demo day. Admittedly, the roughly $100,000 investment Contrary deploys to its companies wouldn’t get your average Silicon Valley startup very far, but for students based in college towns across the U.S., it’s a game-changing deal.

“It gives you a tremendous amount of time to figure things out,” Tarczynski said, noting his own experience building a company while still in school. “We are trying to push them. This is the first time in many cases that these people are working on their companies full-time. This is the first time they are going all in.”

Contrary invests a good amount of its capital in Berkeley, Stanford, Harvard and MIT students, but has made a concerted effort to provide capital to students at underrepresented universities, too. To date, the team has completed three investments in teams out of Stanford, two out of MIT, two out of University of California San Diego and one each at Berekely, BYU, University of Texas-Austin, University of Pennsylvania, Columbia University and University of California Santa Cruz.

“We wanted to have more come from the 40 to 50 schools across the U.S. that have comparable if not better tech curriculums but are underserviced,” Tarczynski explained. “The only difference between Stanford and these others universities is just the volume. The caliber is just as high.”

Contrary’s portfolio includes Memora Health, the provider of productivity software for clinics; Arc, which is building metal 3D-printing technologies to deliver rocket engines; and Deal Engine, a platform for facilitating corporate travel.

“We are one giant talent scout with all these different nodes across the country,” Tarczynski added. “I’ve spent every waking moment of my life the last eight years living and breathing university entrepreneurship … it’s pretty clear to me who is an exceptional university-based founder and who is just caught up in the hype.”

China’s Tesla wannabe Xpeng starts ride-hailing service

in alibaba/Asia/Automotive/Beijing/Cars/carsharing/China/Delhi/didi/Didi Chuxing/electric car/Elon Musk/Foxconn/IDG Capital/India/New York/Politics/serial entrepreneur/shenzhen/Tesla/transport/Transportation/ucweb/Xiaopeng by

There’re a lot of synergies between electric vehicles and ride-hailing. Drivers are able to save more steering an EV compared to a gas vehicle. Environmentally conscious consumers will choose to hire an electric car. And EVs are designed with better compatibility with autonomous driving, which is expected to hit the public road in the coming decades.

Indeed, Tesla is eyeing to launch its first robotaxis in 2020 as part of a broader ride-sharing scheme. Over in China where Tesla has a few disciples, EV startup Xpeng Motors, also known as Xiaopeng, just started offering a ride-hailing app powered by its own electric fleets.

Screenshot of Xpeng’s ride-hailing app ‘Youpeng Chuxing’

The company is the latest in a clutch of carmakers flocking to introduce their own ride-hailing platforms. Didi Chuxing’s massive loss has not deterred their ambitious plans. Rather, this may be a prime time to crack a market long dominated by Didi, which is prioritizing safety over growth following two high-profile incidents and a series of new government regulations.

Xpeng’s ride-hailing app is currently only available in a limited area within Guangzhou where it’s headquartered, shows a test conducted by TechCrunch’s on Thursday.

The company’s coffer is probably large enough to fund its newly minted venture. It’s one of the most-backed EV upstarts alongside rival Nio, which raised $1 billion from a New York initial public offering last year.

Xpeng has to date banked $1.3 billion from Alibaba, IDG Capital, Foxconn, UCAR and other big-name investors, according to disclosed funding data collected by Crunchbase. Founder He Xiaopeng, a serial entrepreneur who made a fortune selling his mobile browser company UCWeb to Alibaba, told CNBC in March that Xpeng may also try an IPO down the road but wants to focus on building the business first.

When it comes to sources of inspiration for the business, Xpeng told local media that it sees Tesla as its “benchmark”. The company has never been shy about its admiration for its American peer. In an interview with Quartz in 2018, He said one of the reasons he founded Xpeng “was because Elon Musk made Tesla’s patents available. It was so exciting.”

But the affection might have gone a little far. In March, Tesla sued an ex-employee for allegedly stealing Autopilot’s proprietary technology before taking a job at Xpeng.

Xpeng started shipping to its first owners in March and was founded five years ago against the backdrop of Beijing’s aggressive electric push in the transportation sector. The sprawling city Shenzhen, just north to Hong Kong, has turned all its public buses and almost all of its taxis electric.

Some reassuring data for those worried unicorns are wrecking the Bay Area

in Airbnb/Apple/austin/california/Column/Companies/CrunchBase/Delhi/Entrepreneurship/Facebook/founder/Fremont/Google/India/Instacart/Intel/Lyft/Nashville/neologisms/palo alto/phoenix/Pinterest/Politics/Portland/Private equity/San Francisco/san francisco bay area/Silicon Valley/Startup company/Startups/TC/Tesla/Uber/unicorn/valuation by

The San Francisco Bay Area is a global powerhouse at launching startups that go on to dominate their industries. For locals, this has long been a blessing and a curse.

On the bright side, the tech startup machine produces well-paid tech jobs and dollars flowing into local economies. On the flip side, it also exacerbates housing scarcity and sky-high living costs.

These issues were top-of-mind long before the unicorn boom: After all, tech giants from Intel to Google to Facebook have been scaling up in Northern California for over four decades. Lately however, the question of how many tech giants the region can sustainably support is getting fresh attention, as Pinterest, Uber and other super-valuable local companies embark on the IPO path.

The worries of techie oversaturation led us at Crunchbase News to take a look at the question: To what extent do tech companies launched and based in the Bay Area continue to grow here? And what portion of employees work elsewhere?

For those agonizing about the inflationary impact of the local unicorn boom, the data offers a bit of reassurance. While companies founded in the Bay Area rarely move their headquarters, their workforces tend to become much more geographically dispersed as they grow.

Headquarters ≠ headcount

Just because a company is based in Northern California doesn’t mean most workers are there also. Headquarters, our survey shows, does not always translate into headcount.

“Headquarters location can often be the wrong benchmark to use to identify where employees are located,” said Steve Cadigan, founder of Cadigan Talent Ventures, a Silicon Valley-based talent consultancy. That’s particularly the case for large tech companies.

Among the largest technology employers in Northern California, Crunchbase News found most have fewer than 25 percent of their full-time employees working in the city where they’re headquartered. We lay out the details for 10 of the most valuable regional tech companies in the chart below.

With the exception of Intel, all of these companies have a double-digit percentage of employees at headquarters, so it’s not as if they’re leaving town. However, if you’re a new hire at Silicon Valley’s most valuable companies, it appears chances are greater that you’ll be based outside of headquarters.

Tesla, meanwhile, is somewhat of a unique case. The company is based in Palo Alto, but doesn’t crack the city’s list of top 10 employers. In nearby Fremont, Calif., however, Tesla is the largest city employer, with roughly 10,000 reportedly working at its auto plant there.(Tesla has about 49,000 employees globally.)

Unicorns flock to San Fran, workers less so

High-valuation private and recently public tech companies can also be pretty dispersed.

Although they tend to have a larger percentage of employees at headquarters than more-established technology giants, the unicorn crowd does like to spread its wings.

Take Uber, the poster child for this trend. Although based in San Francisco, the ride-hailing giant has fewer than one-fourth of its employees there. Out of a global workforce of around 22,300, only about 5,000 are SF-based.

It’s unclear if that kind of breakdown is typical. We had trouble assembling similar geographic employee counts at other Bay Area unicorns, mainly because cities break out numbers only for their 10 largest employers. The lion’s share of regional unicorns are San Francisco-based, and of them only Uber made the Top 10.

That said, there is another, rougher methodology for assessing who works at headquarters: job postings. At a number of the most valuable Bay Area-based unicorns — including Airbnb, Juul, Lime, Instacart, Stripe and the now-public Lyft —  a high number of open positions are far from the home office. And as we wrote last year, private companies have been actively seeking out cities to set up secondary hubs.

Even for earlier-stage startups, it’s not uncommon to set up headquarters in the San Francisco area for access to financing and networking, while doing the bulk of hiring in another location, Cadigan said. The evolution of collaborative work tools has also enabled more companies to add staff working remotely or in secondary offices.

Plus, of course, unicorn startups tend to be national or global in focus, and that necessitates hiring where their customers are located.

Take our jobs, please

As we wrap up, it’s worth bringing up how unusual it once was for denizens of a metro area to oppose a big influx of high-skill jobs. In the past couple of years, however, these attitudes have become more common. Witness Queens residents’ mixed reactions to Amazon’s HQ2 plans. And in San Francisco, a potential surge of newly minted IPO millionaires is causing some consternation among locals, along with jubilation among the realtor crowd.

Just as college towns retain room for new students by graduating older ones, however, it seems reasonable that sustaining Northern California’s strength as a startup hub requires locating jobs out-of-area as companies scale. That could be good news for other cities, including Austin, Phoenix, Nashville, Portland and others, which have emerged as popular secondary locations for fast-growing unicorns.

That said, we’re not predicting near-term contraction in Bay Area tech employment, particularly of the startup variety. The region’s massive entrepreneurial and venture ecosystem keeps on producing valuable newcomers well-capitalized to keep hiring.

Methodology

We looked only at employment at company headquarters (except for Apple) . Companies on the list may have additional employees based in other Northern California cities. For Apple, we included all Silicon Valley employees, per estimates by the Silicon Valley Business Journal.

Numbers are rounded to the nearest hundred for the largest employers. Most of the data is for full-time employees only. Large tech employers hire predominantly full-time for staff positions, so part-time, whether included or not, is expected to reflect only a very small percentage of employment.

Cities list their 10 largest employers in annual reports. We used either the annual reports themselves or data excerpted in Wikipedia, using calendar year 2017 or 2018.

Tesla vaunts creation of ‘the best chip in the world’ for self-driving

in Artificial Intelligence/Automotive/autonomous vehicles/Delhi/India/Politics/robotics/self-driving/self-driving cars/Tesla/Tesla Autonomy Day/Transportation by

At its “Autonomy Day” today, Tesla detailed the new custom chip that will be running the self-driving software in its vehicles. Elon Musk rather peremptorily called it “the best chip in the world…objectively.” That might be a stretch, but it certainly should get the job done.

Called for now the “full self-driving computer,” or FSD Computer, it is a high-performance, special-purpose chip built (by Samsung, in Texas) solely with autonomy and safety in mind. Whether and how it actually outperforms its competitors is not a simple question and we will have to wait for more data and closer analysis to say more.

Former Apple chip engineer Pete Bannon went over the FSDC’s specs, and while the numbers may be important to software engineers working with the platform, what’s more important at a higher level is meeting various requirements specific to self-driving tasks.

Perhaps the most obvious feature catering to AVs is redundancy. The FSDC consists of two duplicate systems right next to each other on one board. This is a significant choice, though hardly unprecedented, simply because splitting the system in two naturally divides its power as well, so if performance were the only metric (if this was a server, for instance) you’d never do it.

Here, however, redundancy means that should an error or damage creep in somehow or another, it will be isolated to one of the two systems and reconciliation software will detect and flag it. Meanwhile the other chip, on its own power and storage systems, should be unaffected. And if something happens that breaks both at the same time, the system architecture is the least of your worries.

Redundancy is a natural choice for AV systems, but it’s made more palatable by the extreme levels of acceleration and specialization that are possible nowadays for neural network-based computing. A regular general-purpose CPU like you have in your laptop will get schooled by a GPU when it comes to graphics-related calculations, and similarly a special compute unit for neural networks will beat even a GPU. As Bannon notes, the vast majority of calculations are a specific math operation and catering to that yields enormous performance benefits.

Pair that with high speed RAM and storage and you have very little in the way of bottlenecks as far as running the most complex parts of the self-driving systems. The resulting performance is impressive, enough to make a proud Musk chime in during the presentation:

“How could it be that Tesla, who has never designed a chip before, would design the best chip in the world? But that is objectively what has occurred. Not best by a small margin, best by a big margin.”

Let’s take this with a grain of salt, as surely engineers from Nvidia, Mobileye, and other self-driving concerns would take issue with the statement on some grounds or another. And even if it is the best chip in the world, there will be a better one in a few months — and regardless, hardware is only as good as the software that runs on it. (Fortunately Tesla has some amazing talent on that side as well.)

(One quick note for a piece of terminology you might not be familiar with: OPs. This is short for operations for second, and it’s measured in the billions and trillions these days. FLOPs is another common term, which means floating-point operations per second; these pertain to higher-precision math often used by supercomputers for scientific calculations. One isn’t better or worse than the other, and they shouldn’t be compared directly or considered exchangeable.)

Update: Right on cue, Nvidia objected to Tesla’s comparison in a statement, calling it “inaccurate.” The Xavier chip Tesla compared its hardware favorably to is a more lightweight chip for autopilot-type features, not full self driving. The 320-TOP Drive AGX Pegasus would have been a better comparison, the company said — though admittedly the Pegasus pulls about four times as much power. So per-watt Tesla comes out ahead by the stats we’ve seen. (Chris here called it during the webcast.)

High-performance computing tasks tend to drain the battery, like doing transcoding or HD video editing on your laptop and it bites the dust after 45 minutes. If your car did that you’d be mad, and rightly so. Fortunately a side effect of acceleration tends to be efficiency.

The whole FSDC runs on about 100 watts (or 50 per compute unit), which is pretty low — it’s not cell phone chip low, but it’s well below what a desktop or high performance laptop would pull, less even than many single GPUs. Some AV-oriented chips draw more, some draw less, but Tesla’s claim is that they’re getting more power per watt than the competition. Again, these claims are difficult to vet immediately considering the closed nature of AV hardware development, but it’s clear that Tesla is at least competitive and may very well beat its competitors on some important metrics.

Two more AV-specific features found on the chip, though not in duplicate (the compute pathways converge at some point), are some CPU lockstep work and a security layer. Lockstep means that it is being very carefully enforced that the timing on these chips is the same, ensuring that they are processing the exact same data at the same time. It would be disastrous if they got out of sync either with each other or with other systems. Everything in AVs depends on very precise timing while minimizing delay, so robust lockstep measures are put in place to keep that straight.

The security section of the chip vets commands and data cryptographically to watch for, essentially, hacking attempts. Like all AV systems, this is a finely-oiled machine and interference must not be allowed for any reason — lives are on the line. So the security piece watches the input and output data carefully to watch for anything suspicious like spoofed visual data (to trick the car into thinking there’s a pedestrian, for instance) to tweaked output data (say to prevent it from taking proper precautions if it does detect a pedestrian).

The most impressive part of all might be that this whole custom chip is backwards-compatible with existing Teslas, able to be dropped right in, and it won’t even cost that much. Exactly how much the system itself costs Tesla, and how much you’ll be charged as a customer — well, that will probably vary. But despite being the “best chip in the world,” this one is relatively affordable.

Part of that might be from going with a 14nm fabrication process rather than the sub-10nm process others have chosen (and to which Tesla may eventually have to migrate). For power savings the smaller the better and as we’ve established, efficiency is the name of the game here.

We’ll know more once there’s a bit more objective — truly objective, apologies to Musk — testing on this chip and its competition. For now just know that Tesla isn’t slacking and the FSD Computer should be more than enough to keep your Model 3 on the road.

1 2 3 18
Go to Top