Search This Blog

Thursday, December 10, 2015

Driving Change: Big Auto’s $2.7 Billion Bet on High-Definition Digital Mapping

As reported by SlashGear:
HERE's plans for the self-driving future can finally emerge from their autopilot holding-pattern, with the company and its new automaker owners holding a quiet but enthusiastic coming out party in San Francisco today. Representatives from Audi, BMW, and Mercedes-Benz parent Daimler joined HERE execs to talk about the development of high-definition mapping, location-based services, and cars that whisper to the cloud and - eventually - drive themselves.
It's the first time representatives from the four companies have sat down publicly and discussed the deal, the collective mouth having been sealed until the ink was dry on Nokia's contract.
"The acquisition of HERE by three competitors in the luxury sector is unprecedented," Klaus Froehlich, member of the board of management of BMW AG, said today.
"Automated driving will become one of the major trends in the automotive sector … it's always a question of when and not if," he pointed out. "HERE is going to provide a high-definition map for us … the resolution will be up as much as 10x compared to today … we need 99.5-percent [accuracy] to be safe on the roads."
here-sg-0
While it may be an acquisition, the deal HERE cut with the consortium of automakers - I'm told there's no official name for the trio - is surprising in its flexibility.
For a start, though they're putting in $2.7bn, Audi, BMW, and Daimler don't walk away with joint control of the mapping specialist. Instead, it'll operate as an independent company, with a new supervising board that will include not only representatives from the automakers but independent nav-industry specialists too.
"HERE will remain open to other investors, and we will invite other partners to join," Froehlich insisted. "We simply wanted to project the company, because when we are realistic about the automotive community we only have three [mapping] providers."
here-sg-1
So, Audi, BMW, and Mercedes-Benz will stay as HERE customers on the one hand, even as their connected cars feed fresh data into HERE's cloud.
Right now, more than two million cars from the three are already cellularly-enabled, and plans are afoot to supply anonymized data from those vehicles that will help both with existing mapping and the new generation of high-definition maps that will be instrumental in developing smart cities and autonomous vehicles.
HERE already gathers what it calls "probe" data from smartphones running its software and other devices, consisting of nothing more than a unique ID, location, speed, and direction. From that - two billion data points a day, in fact - it can figure out traffic conditions as well as identify new roads based on vehicle flow.
The next step to that, HERE's vice president of reality capture John Ristevski explained to me, is baking in data from the variety of extra sensors in the modern car. Such "rich sensor feeds" could provide details on individual lane closures, weather conditions - like whether the windshield wipers are on, or if the traction control is struggling with slippery roads - and even unexpected issues like animals in the path of traffic.
HERE Traffic probe data
"Until now," Froehlich pointed out, "the sensors have only been used for driving assistance services within the car."
It's early days, but what's taking shape is a combination of local processing and cloud services - "in the end," the BMW exec said, "each car only has perhaps 99.9 or 99.8 percent availability of the network" - that would maximize performance while minimizing how much data would have to ping through the mobile networks.
"The amount of data is not as high as one might expect," Dr. Peter Steiner, head of Audi electronics venture, explained. "We are not video streaming, we are feeding back or broadcasting what is already a condensed data set."
So, onboard processing would handle basic feature-extraction, transmitting the core details to HERE's cloud, while anything more complex or confusing to the car could be sent as images for the more advanced machine learning to figure out.
self-driving-mercedes-f-015-1280x632
Privacy is the looming specter around any connected car service, and it's something all four companies are taking seriously from the outset. Mindful, perhaps, that even the whiff of mishandled data could set smart mobility back decades, the consortium is taking a hard line.
"Privacy will become more important, more premium in the future," BMW's Froehlich argues, "we do not accept partners that compromise the privacy of our customers."
"Data privacy, this is the utmost thing for us," Sajjad Khan, VP of digital vehicle and mobility at Daimler, concurred, "next to data security, which is just as competitive."

True autonomy is still some way out. Froehlich envisages it taking two to three years before a standard for dynamically-updated and maintained mapping data is settled upon, which takes into account sensor information from individual vehicles.
"I think in certain regions, automated driving will be even faster," he predicted, such as ride-sharing services where cars would only need to have HD mapping for a single city or location. For full autonomy, though, "I don't see it within the next eight to ten years, honestly," Froehlich says.
That may seem conservative, but it means full speed ahead for HERE. "We do not have much time to waste," he concluded, "we need this new generation of maps as soon as possible, because the gradient of driving to assisted or autonomous driving is quite steep."

Wednesday, December 9, 2015

Tiny IoT Temperature Sensor Powered Wirelessly with Radio Waves

As reported by GizMagOne of the problems for the smart buildings of tomorrow is that they may depend on some very un-smart wires to power them. To cut the cord, Eindhoven University of Technology (TU/e) researcher Hao Gao, as part of his PhD thesis, has developed a tiny transmitting temperature sensor that is powered by radio waves, eliminating the need for wires or batteries. Instead, it picks up radio waves from a special router, converts them into electricity, and uses it to transmit readings.

Like many other forecasters, TU/e sees a future where smart buildings are filled with sensors and other devices that gather information and carry out tasks to automate the business of living while making it more sustainable. But as sensors become smaller and are incorporated into more things around the house, the problem of powering them without stringing tiny wires everywhere or spending half the time swapping out batteries remains.

Part of the PREMISS project, Gao's sensor is very much in the lightweight category at 2 sq mm and weighing in at 1.6 mg. Its operating principle is similar to that of the Thing, which is infamous in espionage circles as the Soviet bug that listened in at the US embassy in Moscow for about six years.
In 1946, a carved replica of the Great Seal of the United States was presented by the Soviets to the embassy as a goodwill gift that seemed the picture of innocence. What that Americans didn't know is that under the woodwork was the Thing, which was a listening device with a monopole antenna. Instead of running off batteries or mains current, the Thing was energized by a radio beam directed at it by the KGB and transmitted back when they wanted to eavesdrop on the ambassador. This meant that it could operate indefinitely and was only discovered in 1952 by accident when a British radio operator tuned in on its transmissions.
Various research teams have also been examining the potential of scavenging ambient radio waves to power all manner of devices, such as biomedical implants and smartphones
Based on 65-nm CMOS technology, the new thermometer system uses a special router that targets the sensor, but unlike the Thing the reason is to save power rather than escape detection. At the same time, the sensor itself is designed to use very little electricity. When the sensor is exposed to radio waves, it absorbs the energy until it stores enough to transmit a signal back to the router. The temperature of the sensor alters the frequency of the signal, which the router can decode.
According to TU/e, the tiny size and independent nature of the sensor means that it can be placed in all sorts of unorthodox locations, such as in plaster or concrete. It can even be mixed in with latex and applied directly to walls like paint.
The current version of the sensor has an operating range of only 2.5 cm (1 in), but it's hoped that this will be extended to a meter (3 ft) in a year and ultimately to 5 m (16 ft). With a mass production cost projected at about 20 cents, TU/e sees a wide range of applications for the wireless sensors, including payment systems, wireless ID, smart buildings, and industrial applications.
Gao's PhD thesis can be found here.

Tuesday, December 8, 2015

Nanomaterial for GPS Clocks Most Expensive in the World at $150 Million per Gram

As reported by the IndependentOxford University scientists are creating the world's most expensive material - endohedral fullerenes, spherical carbon molecules containing nitrogen atoms, which sell for £100 million ($150 million) a gram.

The tiny structures are being manufactured by Designer Carbon Materials, a company which was born out of the university last year.
The incredibly valuable material is being used in atomic clocks, to make the timekeeping deviceseven more accurate than ever before.
When integrated into a GPS device, the tiny clocks could detect the device's position to an accuracy of one millimeter, compared to the current standard of around one to five meters.
That accuracy would be practically unnoticeable if you were navigating a city with Google Maps, but it's vital in driverless car technology, where the difference between meters and millimeters is hugely important to avoid collisions.
Speaking to The Telegraph, Dr Kyriakos Porfyrakis, founder of the company and nano-materials expert, said: "Imagine a miniaturized atomic clock that you could carry around in your smartphone."
"This is the next revolution for mobile."
Most current atomic clocks are large, in some cases, cabinet-sized devices. Using the endeohedral fullerene technology, they could be shrunk to the size of a microchip.
Because of the enormous price, the material changes hands in tiny quantities.
The company recently made their first sale of only 200 micro-grams - about one-fifteenth the weight of a snowflake, or one-third the weight of a single human hair - for £22,000 ($33,000).

Sunday, December 6, 2015

Success! Cygnus Spaceship Launch Restarts Orbital ATK Cargo Missions for NASA

As reported by Space.com: With a brilliant afternoon launch, the private spaceflight company Orbital ATK returned its Cygnus cargo ship to flight after a year on hiatus Sunday (Dec. 6), launching vital supplies and NASA gear to the International Space Station.

Saturday, December 5, 2015

Porsche Is Charging Ahead With Mission E 'Tesla Killer' Electric Car

As reported by ForbesEven Porsche is gunning for Tesla Motors these days. After floating the Mission E pure electric concept in September, Porsche is going ahead with production of the vehicle.

“The first 100% electrically powered Porsche is on its way. It will be launched at the end of the decade,” Porsche said Friday in a statement. Some highlights from the announcement:
Unveiled at the Frankfurt Motor Show, the four-door car with four individual seats has a system power output of over 440 kW (600 PS). The vehicle will thus achieve both acceleration of 0 to 100 km/h in under 3.5 seconds and a range of more than 500 kilometres. 
With the Mission E project, Porsche is continuing to back sustainable growth. In Stuttgart-Zuffenhausen alone more than 1,000 new jobs are being created. The company will be investing around 700 million euros in its main site there. Over the next few years, a new paint shop and a new assembly plant will be built. The existing engine factory is also being expanded for the production of electric motors. In addition, the existing body shop is being enlarged. On top of that come other areas in which the company will be investing in this context, such as in the Weissach development centre.
Other goodies include: an 800-volt charger that juices up the battery to 80 percent of the range in about 15 minutes. And the vehicle can optionally be recharged wirelessly by induction via a coil set into the garage floor, Porsche said.
The giant asterisk to all of the above is of course that Porsche is doing this by the end of the decade. The Tesla Model S is here and now.

Friday, December 4, 2015

Pro Drivers Go Headset-to-Headset in VR Race on Two Separate Tracks (Video)

As reported by The Wall Street JournalIn a race that fuses video-game technology and real world driving skill, two professional drivers, on two separate but identical tracks, have raced against each other while wearing virtual reality headsets attached to their crash helmets.
The race, which took place on November 10 at Los Angeles’ Santa Anita track, saw the drivers looking not at the actual road in front of them, but at a virtual reality representation of the track, infused with smoke, fire, waterfalls, and boulders.
The drivers, one of whom drove a stunt car in the latest James Bond film ‘Spectre,’ hurtled around the circuit in two identical 2015 V8 Ford Mustangs, trusting that what they were seeing on their Oculus Rift DK2 VR headsets was a true, real-time representation of how their cars were performing on the actual track.
“The main risk is, if I go blind and I can’t see where I’m going at 70 mph, I could thunder into a tree which would not feel very nice,” said U.K.-based stunt driver Ben Collins.
“The beautiful thing about virtual reality is you can press reset without killing yourself.”
To get the real-virtual mix right, technologists working for car oil company Castrol, which produced the event, mixed virtual reality with self-driving car technology.
One of the main challenges: tracking the cars’ exact positions as they sped around the track without the need for re-calibration. This was necessary so that an exact match could be achieved between what was happening on the physical track and its representation on the VR screens.
To track the cars, a system frequently used to navigate robots around a space was used, integrating a combination of physical, optical, and laser sensors gauged the position of the cars at high speeds during the race.
To get the two cars to communicate their locations without any significant transmission latency within the headset–which could cause the drivers to crash as they would not be accurately judging their vehicles’ orientation and locations– the technology team had to make sure that a fast Internet connection in the area of the race tracks would send all the data from the cars and the surrounding cameras and sensors to a powerful, on-site server.
The race comes as VR hardware and software is expected to hit the mainstream market in 2016, with companies like Facebook Inc. marketing the Oculus Rift VR headset and Sony Corp. promoting its PlayStation VR.

Thursday, December 3, 2015

Google's New Computer 'Vision' Tech is Very Cool and Somewhat Terrifying

As reported by Business InsiderGoogle on Tuesday announced a new tool on Wednesday for app developers that's absolutely amazing and also a little disturbing: it lets apps, robots and drones "see."

The tech is called the Google Cloud Vision API and it allows any app developer to tap into Google's smart "machine learning" service that can identify objects, including faces and emotions.
Cloud Vision solves a hard computer problem of "seeing." For instance, your computer can scan and reproduce an image of a mountain or an image of a baby, but to the computer, they look the same: a bunch of pixels. Your computer can't sift through your photos and find baby photos for you, unless you've tagged them "baby."
Google Photos, on the other hand, can find the baby photos.
And now Google is making that technology available to programmers to add to their apps. 
Cloud Vision can even detect various emotions on a face such as a happy smile or an angry frown, Google says.
To demonstrate the power of Cloud Vision, Google built a cute little robot with Cloud Vision, via a DYI robot kit known asGoPiGo. 
At 1:14 in the video below, the robot demonstrates how it can follow faces and detect emotions. At 2:08, the robot demonstrates how it can detect and identify other objects like glasses, a banana, money.
As for the terrifying part ...
While the demo robot is 'adorbs', it doesn't take much imagination to see less cuddly uses for computer vision.
For instance, at least one drone maker has been testing Google Vision out: Aerosense, owned by Sony Mobile Communications. 
"We have drones that take thousands of photos per flight. We find that Google Cloud Vision API as the best way to turn those huge number of photos, automatically produced, into meaningful insights,"Aerosense General Manager Tomoaki Kobayakawa says on Google's blog.
And given Amazon's recent news that its drone delivery project is progressing, we can't help thinking of that Audi commercial featuring people-seeking delivery drones run-amok.
Drones Audi commercial