Ford working with Bloomberg Aspen Initiative on Cities and Autonomous Vehicles
Hyundai adds Blue Link skill for Amazon Alexa

Intel to invest more than $250M over next two years in autonomous driving; “Data is the new oil”

In a keynote address at the AutoMobility LA conference, Intel CEO Brian Krzanich announced that Intel Capital is targeting more than $250 million of additional new investments over the next two years to make fully autonomous driving a reality. This is the first time Intel is keynoting at an automotive conference, signifying how critical the automotive market has become for the company.

These investments will drive the development of technologies that push the boundaries on next-generation connectivity, communication, context awareness, deep learning, security, safety and more. Drilling down into the areas that will be fueled by the fresh investments, Krzanich highlighted technologies that will drive global Internet of Things (IoT) innovation in transportation; areas where technology can directly mitigate risks while improving safety, mobility, and efficiency at a reduced cost; and companies that harness the value of the data to improve reliability of automated driving systems.

When it comes to the car of the future and automated driving experiences, data is literally the new oil. Data has the potential to radically change the way we think about the driving experience: as consumers, as automakers, as technologists and as citizens of our communities. In fact, as a technologist, one of the trends I see as most disruptive to almost every industry is the enormous flood of data driven by the proliferation of smart, connected devices.

—Brian Krzanich

IMG_0339

In his talk, Krzanich noted that in 2016, the average person generates 650MB of data a day through use of their PCs, mobile phones and wearables. By 2020, projections show that the average person will generate 1.5GB of data a day. Despite the impressive ramp up, however, that volume of data is paltry compared to what will be generated by autonomous vehicles.

Each autonomous vehicle, supported by it platoon of sensors—cameras, radar, sonar, GPS and LiDAR—will be generating approximately 4,000 GB—i.e., 4 terabytes—of data per day, Krzanich said. Cameras will generate 20-60 MB/second, radar upwards of 10 kB/s, sonar 10-100 kB/s, GPS will run at 50 kB/s, and LIDAR will range between 10-70 MB/s.

Every autonomous car will generate the data equivalent of almost 3,000 people. Extrapolate this further and think about how many cars are on the road. Let’s estimate just 1 million autonomous cars worldwide—that means automated driving will be representative of the data of 3 billion people. Just as oil has transformed our world over the last century, data is poised to transform our world for the next hundred years – and beyond.

—Brian Krzanich

In his talk, Krzanich differentiated between three types of data generated and used by autonomous vehicles:

  • Technical data. This data comes from the car’s sensors that interpret the difference between a child or an animal, a fallen branch or a traffic cone and directs the outward decisions and movements of the car. This data takes an incredible amount of compute power, and whoever has the best data can develop the best artificial intelligence tools of machine learning, deep learning algorithms and data analytics.

  • Societal and crowdsourced data. Data from the world around the vehicle, such as traffic, influences how the car gets from point A to point B, and how the car can change course to point C. The Waze app is a good example of this type of data today. With this data, whoever has the most data will be able to develop the best applications.

  • Personal data. Data that tracks how many people are in the car, music preferences of each passenger, or even what stores or brands passengers prefer and, when you are near them, tees up sale items. Wearables and other sensors inside the car can also monitor behavior, focus, emotional and biometric status to increase safety and security. Whoever has the most personal data will be able to develop and deliver the best user experience.

Data is truly the new currency of the automotive world. It’s not enough just to capture the data. We have to turn the data into an actionable set of insights to get the full value out of it. To do that requires an end-to-end computing solution from the car through the network and to the cloud – and strong connectivity.

—Brian Krzanich

In addition to the requirement for copious, low-cost, non-volatile memory; field programmable gate arrays (FPGAs); and cloud systems, Krzanich said, 5G communications will be fundamental to realizing the promise of autonomous vehicles.

Comments

HarveyD

If INTEL (and many others) spend all those $$$M in the next 2 or 3 years to develop ADVs programs, Henrik's dream may come true sometime between 2020 and 2022?

Account Deleted

Harvey you can see for yourself at Tesla’s website that Tesla is already selling fully driverless capable cars. All cars that Tesla makes are capable of driverless driving. They only need a software update and to go through a validation and authorization process for any Tesla car to become a fully driverless taxi on the Tesla Network. All this is done by 2018. The rest of the industry will follow by 2020 to 2022 and those who do not will bankrupt shortly thereafter because there will be no demand for non-driverless cars.

By dec 2020 Tesla could have over 2 million driverless taxis on the Tesla Network each with the ability to earn 1 USD per mile driven. Many if not most of these cars could do 100.000 miles per year and thus earn 100.000 USD per year in taxi fees. Each million Tesla taxies could therefore make 100 billion USD per year in taxi fees and Tesla. The cost to do that will only be about 20 billion USD so we are talking about 80 billion USD in profits for Tesla and Tesla’s vehicle owners. Tesla will be done asking for capital in the stock market or anywhere else when they launch the Tesla Network. They will have all the money needed to expand production to as many million cars per year as they can find demand for and do it very fast. Tesla could be making 10 million cars per year by 2025 with gigafactories all over the planet if they do not get serious competition.

Trees

The EV should own the Metro market for transportation. They can charge up every 100 miles and displace pollution to countryside. My guess they would displace public transportation and make for efficient and convenient city transportation. The car will become basic utility grade as non owner occupied vehicle is designed for low cost, comfort, and convenience. A easy to clean vehicle as the public can be very dirty, (garbage and fluids). I can't see the public paying $1/mile for such a transport, given the competition will be fierce. My guess the cost will spiral south and make ownership of cars a bad decision. Vehicle sales of ICE will continue. Trucking, RV, towing, off road, remote, long distance travel, and emergency/back up will continue to utilize liquid fuels per the low cost and convenience. Biofuels will continue to gain popularity per lower cost and environmental concerns.

Account Deleted

I think public transportation will still be there. But it will have to evolve to compete with taxi services such as the Tesla Network. You could travel from NYC to LA in 4 hours using a hyperloop. City center to city center, no noise and no air pollution. Also metro stations underground is not going away. Large cities do not have enough surface roads to handle the traffic so we need underground metros.

I also thought about the cleaning issue. Autonomous taxies will have to visually monitor its passenger(s) and fine them for any destruction or extraordinary cleaning that their travel causes. So I think people will behave much nicer than in traditional non-monitored public transportation. The Tesla Network has the identity and credit card info of each customer so it should not be difficult to make an effective system preventing abuse of the vehicles. Tesla taxis that are owned by private people will probably be cleaned by the owners or the owner will pay for a Tesla service to get it cleaned when needed.

The world’s first full scale hyperloop is now being build see here https://electrek.co/2016/11/15/hyperloop-one-first-tube-devloop-las-vegas/

mahonj

It has taken Intel quite a while to cotton on to this.
I would say NVIDIA has got the march on them.
@Henrik,
Having a car that is AV ready is not at all the same as having one that is operating safely in all reasonable conditions.
The hardware is relatively easy (much of the heavy work being done by the mobile phone world), the s/w is the hard part.

Account Deleted

Being hard is still not stopping Tesla's programmers from being ready with the fully self-driving software by dec2017. Musk has said a 100% hands free test drive from LA to NYC by dec 2017 will show just how powerful the system is. You will see an endless stream of software upgrades to autopilot 8 starting Dec 2016 and culminating with the real stuff in Dec 2017. After that it is further improvement and data gathering. By Dec 2018 Tesla Network will be operational in most of USA I expect.

Account Deleted

NVIDIA is "lucky" that AI computations are very suited for GPUs that NVIDIA excels in. Intel is a CPU expert but they can not be used so much for AI computations. I think this is why NVIDIA is ahead.

Trees

Companies will have to work together to generate standards. The litigation and safety issue is to high to have competing technology. Software will have to be bullet proof or the fed gov't will take over per politics.

Point to point public transportation is highly desirable. Mass transit is littered with Union workforce
management problems, terrorism/security concerns, high taxpayer costs, passenger safety, inflexibility, miss management and political chicanery. International travel will utilize air transport, but almost all public transport by bus or train would go obsolete. I do think NYC could best be served by efficient autonomous vehicle network 24x7. Cross walks may need to be above or below grade and high speed single lane traffic at street center with off and on electronic ramping.

SJC

Tesla autopilot: "98% of those surveyed said they understand that the driver must maintain control.."

http://insideevs.com/tesla-autopilot-survey-results-98-understand-that-autopilot-requires-driver-attention/

Some people STILL don't understand this.

Account Deleted

You do not need standards that will only slow down or prevent innovation. You do need data to scientifically confirm that the self driving AI is better than humans to do the job. Auto-makers must be able to prove on a continuous basis that their auto pilot is importantly safer than human drivers or they will have their license for public use of self driving taxis withdrawn.

Air transport is not sustainable and I see little other way than hyperloops to take over. Even if we create synthetic co2 neutral fuel they still pollute when burned in a jet engine. We could use liquid hydrogen but it will be very expensive as the number of passengers per plane will be small like 50 passengers for a 747 sized planes and the h2 fuel is more expensive. I think hyperloops are the cheapest way to go. It is also supersonic and all weather capable.

mahonj

@Trees, you can't really beat public transport (buses, trams, trains) for densely populated cities. 80 people on a bus take the same space as 2 cars, which usually carry 2 people.
(They could carry 8 people, but you almost never see this, especially at rush hour.)
You are right to complain about unionised drivers but IMO it is a price worth paying. There isn't enough road space to put everyone in cars (electric or ICE, human or AI driven), they all take about the same amount of road space.

Account Deleted

Tesla will build a small self-driving bus build on the Model X that can take 6 people in comfort. It will be filled because a data center will syncronise traffic between this bus and people transported by the smaller Model 3. The plan is the bus will do standard routes back and forth and model 3 taxis will take people the last miles to the dor.

mahonj

So, 12 people in the space of 80, only 7x worse.
Could work well outside rush hour.

Account Deleted

You could squeeze 9 into the model x bus. Three rows with 3 seats. I still think if traffic is that bad in rush hour an underground metro system is better. You can have several layers of tubes with trains in it. Large busses does not follow the traffic they drive slower and accelerate slower. A model x bus with 9 people can flow with the traffic of the other cars. That will make it faster. They may also be able to platoon because they are driverless and react faster than humans.

Account Deleted

I think the model x based minibus will come as soon as Tesla has their self driving pilot approved for level 5 driving. So probably late 2018.

Dr. Strange Love

LOL ... "Data is the new Oil". Tell this to Warren Buffet.

Trees

I think most are missing the efficiency of computer control of auto's. Yes, in the short run a human needs to oversee, but if were talking of future, I don't think humans could compete. Traffic efficiency would be a exponential, and more akin to pipeline efficiency. No slow drivers, no confused drivers, and bumper to bumper traffic traveling fast and upon all roadways. It's really not that hard to be public transport. Also, we must think in terms of convenience. It's not good enough to force residents to slowly walk miles to wait in line and once within a train car to be whisked to a terminal where once again the slowest transport known to man (walking) required to make public transportation possible. We need more productivity within the country. I don't want citizens wasting time. Sure, it's good to get exercise, just should be on your terms.

mike96

Good Post. Please keep updating your posts on daily basis. Users like me are eagerly waiting for them.

Thanks
Mike Craig
[url=http://socialmediafans.wordpress.com/]IT Solutions[/url]

HarveyD

Too many AD cars would create huge traffic jams.

A better solution may be AD e-mini-buses with 10 to 14 passengers. Fully automated e-mini-buses would quickly learn the hour by hour traffic density. Proper size AD e-buses would be automatically selected.

The idea is to reduce the number of vehicles on our street/roads, not to increase their number.

SJC

Data is NOT the "new oil".
Try putting data in your gas tank, see how far you get.

HarveyD

Clean lower cost REs (with battery/H2 storage) will feed future EVs and FCEVs and are the real new OIL.

Data will help to provide large and small EVs/FCEVs with automated driverless capabilities to make them saver and much more convenient.

The comments to this entry are closed.