TomTom « Terug naar discussie overzicht

De hype cycle van self-driving cars

3.904 Posts, Pagina: « 1 2 3 4 5 6 ... 93 94 95 96 97 98 99 100 101 102 103 ... 192 193 194 195 196 » | Laatste
[verwijderd]
0
quote:

*Justin* schreef op 6 september 2016 20:01:

Kijk nog even in herinnering. De wijze waarop nieuwkomers localisatie willen doen was al lang geleden door tomtom geprobeerd en onvoldoende stabiel bevonden. www.economist.com/news/science-and-te...

en dan voor degene die nog even willen doorlezen en willen weten hoe de halve academische wereld hierover denkt:

iv2016.org/wp-content/uploads/2016/06...

Intelligent Vehicle Symposium in Juni 2016
veel papers over algorithmes voor localisation, ook veel alternatieven voor Road DNA, maar wel overeenstemming over een map component matching a visual/radar component.

Mijn conclusies (zonder een echte deep-studie):

Daimler en VW en de duitse academici zijn het wel eens met TT, of een variant hierop.
Shanghai University heeft een paper over RoadDNA, en RoadDNA geeft vooral het voordeel dat het veel minder resources vraagt.
Tokyo en Toyota lijken meer met 3D camera's te doen.
[verwijderd]
0
quote:

Oeps...! schreef op 6 september 2016 20:35:

Volvo and Autoliv are creating a new company to sell self-driving car software

By Lulu Chang — September 6, 2016 8:46 AM

A couple months apparently goes a long way in changing Volvo’s mind when it comes to autonomous cars. Back in July, CEO Håkan Samuelsson made clear his company’s position on driverless technology, noting that Volvo had “no interest in manufacturing or selling fully autonomous cars with no driver involvement.” But now, the Sweden-based firm is partnering with Autoliv Inc. “to set up a new jointly owned company to develop next generation autonomous driving software.” So maybe some things can change.

As the two companies noted in a joint release on Tuesday, the new venture will be headquartered in Gothenburg, Sweden, and will begin its operations in 2017. The plan is for the yet-to-be-named company to “develop advanced driver assistance systems (ADAS) and autonomous drive (AD) systems for use in Volvo cars and for sale exclusively by Autoliv to all car makers globally.” As self-driving cars continue to trend across both automotive and technology companies’ road maps, Volvo and Autoliv are hoping to make their mark on the safety of this next-generation technology.
“By combining our know how and resources we will create a world leader in AD software development,” said Samuelsson, highlighting Volvo’s reputation for making safe, dependable vehicles (the company actually invented the three-point safety belt). “This means we can introduce this exciting technology to our customers faster.”

Jan Carlson, chairman, chief executive and president of Autoliv, echoed these sentiments, noting, “There are no two companies that can claim to have done more for automotive safety worldwide than Autoliv and Volvo. This new company is a recognition of the fact that autonomous driving is the next step to transform road safety.”

Volvo previously announced a partnership with Uber to develop autonomous cars, and has also previously announced the launch of self-driving pilot projects in a number of cities across the world, including Pittsburgh, London, and China.

www.digitaltrends.com/cars/volvo-auto...

Heeft dit gevolgen voor TomTom?
Lijkt mij zeer positief voor TomTom:

-TomTom HD-maps + RoadDNA is ook extra safety.
Sluit dus perfect aan bij deze software-stap van Volvo en biedt op termijn extra afzetmogelijkheden.

-Volvo heeft zeer strategisch gekozen voor TomTom-inside in alle nieuwe Volvo's vanaf 2019. (Vergeet ook niet dat Volvo de meest trouwe klant van Here was. Toch koos men voor de voorsprong en unieke technologie die TomTom kan bieden.)
[verwijderd]
0
quote:

seren schreef op 6 september 2016 21:05:

[...]

en dan voor degene die nog even willen doorlezen en willen weten hoe de halve academische wereld hierover denkt:

iv2016.org/wp-content/uploads/2016/06...

Intelligent Vehicle Symposium in Juni 2016
veel papers over algorithmes voor localisation, ook veel alternatieven voor Road DNA, maar wel overeenstemming over een map component matching a visual/radar component.

Mijn conclusies (zonder een echte deep-studie):

Daimler en VW en de duitse academici zijn het wel eens met TT, of een variant hierop.
Shanghai University heeft een paper over RoadDNA, en RoadDNA geeft vooral het voordeel dat het veel minder resources vraagt.
Tokyo en Toyota lijken meer met 3D camera's te doen.

Het is maar de halve academische wereld. Vooralsnog wedstrijd onbeslist en 50%-50% AI versus HD-Maps? ;-)
[verwijderd]
0
Ik ben op een artikel gestuit over Tesla, een review in de UK en daar hebben ze het over Navigon.
Is dat een "soort" concurrent van TomTom?

Tesla Model S (2016) review: The ultimate electric car in the UK

This might sound like overkill, but in practice, the Tesla has one of the most intuitive, informative and usable navigation systems I’ve used. Short-term directions and lane guidance are displayed in sharp, crisp detail to the left of the dials (oddly, this display uses technology licensed from Navigon, not Google Maps), while you get a better overview of the route on the touchscreen.

These directions are given well before you need them, giving you time to plan ahead, and the only shortcoming is a lack of multi-point route planning.Score: 4.5/5

www.alphr.com/tesla-model-s/1003282/t...
Dutchy Ron
1
What Self-Driving Cars Will Mean for Automakers’ Stocks

fortune.com/2016/09/06/auto-stocks-te...

So are auto stocks doomed to the same old repetitive commute from peak to trough? Maybe not. As cars become the epicenter of innovation, there’s a factor in play this time around that was absent the last time carmakers struggled. That, of course, is the rise of autonomous-driving technology, which could upend the business model of the industry in the not-too-distant future—in ways that carmakers and their investors could profit from. While it’s far too early to pick winners in the self-driving-car arms race, the current down cycle in auto stocks could let investors place some early bets at bargain prices

Geldt ook voor TT ;-)
Dutchy Ron
0
Driverless Vehicles: Ways to Improve Safety and Accelerate Environmental Progress Together

www.huffingtonpost.com/howard-learner...

Federal oversight agencies are initially focusing on autonomous vehicle safety, but they can do more than one thing at a time. Consistent with the Obama Administration’s overall climate change policy solutions, the National Highway Traffic Safety Administration (NHTSA) and the U.S. Environmental Protection Agency (EPA) should assess ways in which autonomous vehicle use can reduce greenhouse gas pollution. This should be a “both-and”: the technological advances that are spurring development of autonomous vehicles can create opportunities to both reduce fatalities on roadways and reduce pollution that harms our health and environment.

Voor de echte autoliefhebbers in de bijlage nog een lijvig rapport (geen directe relatie met TT) (https://www3.epa.gov/otaq/climate/documents/mte/420d16900.pdf)
Dutchy Ron
0
www.recode.net/2016/9/6/12772862/self...

This self-driving startup wants to change the way robot cars make decisions and communicate

From consumers to the experts, there’s still a lot of uncertainty about self-driving cars.

Is the public ready for two-ton automobiles driven by software? Is Lidar — an expensive and often unwieldily technology used to calculate distances using lasers — even necessary? (Elon Musk says no.) And how will self-driving cars communicate with other cars and pedestrians?

This new self-driving startup — like many others before it — purports to have the answers. Drive.ai, co-founded by Carol Reiley and Sameep Tandon, came out of stealth just last week with one basic premise: The answer to all these questions lies within deep learning.

What’s the difference between deep learning and other machine learning systems? At the risk of oversimplifying it, deep learning is a machine learning system that is mirrored closely after the human brain, whereas other machine learning systems often require rules and human programming.



So why do Reiley and Tandon think deep learning is the key to unlocking the true potential of self-driving cars?

Consider this: If you had to give a machine a set of rules that would help it recognize a chair, what would those rules be? Maybe: This chair stands on four legs and is brown.

But what happens when you flip that chair upside down? Or if you introduce a different colored chair? Based on the given rules, that machine may not be able to recognize the object.



But give a person a chair — just a chair — and they’ll be able to recognize chairs everywhere.

Unlike other computer vision and machine learning systems, deep learning systems can roughly think and make decisions like a human brain.

“Other systems use a lot of rules, but autonomous driving is a hard and complicated problem,” Tandon, who is also Drive.ai’s CEO, told Recode. “Deep learning systems are the most effective in a practical, real-world application because it can capture more nuance and different variations than other machine learning algorithms.”

After launching out of stealth early last week, Drive.ai joined the growing fray of self-driving startups with plans to integrate with commercial fleets in an effort to acquire more driving data. Eventually, like the self-driving startups before them, Reiley and Tandon hope to work directly with automakers and have already secured a few fleet partners, though they declined to say who they were.

Drive.ai isn’t alone in its ambitions. Many of the freshman class of self-driving startups that have recently launched fall into one of two buckets: They have either been created in the hope of eventually being acquired by an automaker or, like Drive.ai, they want to be vehicle-agnostic and hope to avoid being restricted to a single legacy automaker.

Think Cruise, which was acquired by General Motors, versus MobileEye, which has become an industry standard. But as the self-driving market gets more competitive, there may be a shift away from automakers working with technology companies that are standard across the industry in order to secure a competitive advantage over its rivals.

For now, the team is focused on its commercial fleet partnerships in order to ensure their technology is safe and reliable. Tandon, who was studying artificial intelligence at Stanford along with several other of the team’s engineers before halting his studies to start Drive.ai, and Johns Hopkins Ph.D. candidate Reiley have so far raised $12 million from undisclosed investors and have a total of 20 employees.

The company also announced that General Motors veteran Steve Girsky joined its board — a move that may bolster its relationships with carmakers (or even hint at an existing one).

Admittedly, Drive.ai isn’t the only company using deep learning algorithms in their self-driving cars, Tandon said. Mobileye, for example, recently announced they would use deep learning for their camera systems.

But Drive.ai is the only one using deep learning to drive the entire system, he said. That means deep learning is powering everything — from the sensors and cameras, to the vehicle’s decision-making, to the way the car communicates with people and things around it.

Communication, like deep learning, is crucial, according to Reiley and Tandon.

Cars, regardless of what system they’re operating, will need to be able to communicate with one another just as people will need to be able to understand the cars’ intentions. As the self-driving space becomes increasingly crowded with the entrance of legacy and new transportation players, standardizing that communication becomes trickier.

Communication, according to Tandon, is also an important method of gaining the trust of the public. After all, consumer trust — like the reliability and legislative approval of the technology — can either be the ultimate impediment to the deployment of self-driving or it can be what expedites it.

“We believe [a standard] needs to exist,” Tandon said. “Cars today have break lights, horns, rear-view mirrors, side-view mirrors — but all those standards developed over time. Our belief with the self-driving car is that a similar transformation will have to occur. For example, we may have to replace the human hand signals, like the wave.”

Already, Tandon and his team have been discussing implementing a standard method of communication with his existing partners.

“We’re still trying to figure out what that standard even is,” he said. “But to me, it’s just a fundamental part of the problem of self-driving cars.”

To that end, the company’s “kit,” which can be retrofitted to existing vehicles, includes LED signs that sit atop the car and speakers that relay the car’s intentions to pedestrians and human drivers around it. (Naturally, the kit is also equipped with sensors like radar, GPS, cameras and Lidar, though Reiley said the team considers Lidar to be more of a redundancy than a necessary part of the system.)

The company is still testing different methods of communication to measure things like how quickly and how well people respond to the audio-visual cues. For example, do pedestrians notice the LED sign that says that it’s safe to cross, or do they need an extra sound or voice?

Using deep learning to power that part of the system, Tandon said, allows it to learn how to communicate in a way that a pedestrian and another human driver might understand
holenbeer
0
quote:

Dokus_ schreef op 6 september 2016 12:55:

[...]

ik zie nvidia ook als grote bedreiging voor tomtom op de langere termijn. Lidar, sensoren worden allemaal kleiner/beter/sneller e vooral betaalbaarder. De px2 is nu al in staat om realtime 3d kaarten te maken... laat staat over 2 jaar met nog betere processoren, nog betere sensoren, nog betere lidar, etc

hdmaps worden dan on the fly in de auto gecreeerd.
Aandelen of opties nvidia zijn te koop, ook die van andere kanshebbers. Ik heb mijn SDC- portefeuille flink gespreid, dan is de kans groter dat ik er een winnaar tussen heb zitten. Hoe kansrijk Tom Tom ook is, beetje spreiden kan geen kwaad
[verwijderd]
0
Navigon lijkt een duits/Garmin software/navigation/app bouwer te zijn, voor Tesla bovenop Google maps. Niet iedereen is blij met die navigatie.

Onafhankelijk hier van toch maar weer wat Tesla gekocht, het is weer tijd voor goed nieuws, Musk die drie dagen stil is, dat moet wat worden. Dat doet ie anders nooit.
Dutchy Ron
0
Aardig artikel over issues als wetgeving, beveiliging en eigenaarschap data, etc.

www.ipwatchdog.com/2016/09/05/privacy...

Driverless cars offer great promise in improving safety and convenience. But, they also pose considerable privacy and security risks. Federal and state governments need to consider all the risks very carefully and make a concerted effort to address them before urging adoption of this revolutionary, new technology

Techniek is niet het probleem of de vertragende factor het gaat om de randvoorwaarden.
Dutchy Ron
0
Gevolgen voor de werkgelegenheid zijn ook fors, mooi moment om deze revolutie te combineren met het basisinkomen (vanaf 2017 beproefd in Finland) ;-)

www.sfchronicle.com/business/article/...
Dutchy Ron
0
Tot slot nog een artikel over draagvlak en acceptatiegraad, 48% wil zelf blijven rijden.

While 94 percent of those surveyed are aware autonomous autos are coming, almost 48 percent aren’t interested in buying one, MaritzCX found. Even among Mercedes and Infiniti owners, more reject than accept the technology. The findings support other recent surveys that show most motorists still aren’t ready to embrace cars that drive themselves.

alabamanewscenter.com/2016/09/05/merc...
Dutchy Ron
0
En nog een artikel over Telematics

www.information-age.com/technology/ap...

Citaat (deel vh artikel):
So, the awareness around connectivity will increase as cars become connected, and many more things become connected in an IoT sense.

Telematics will play a crucial role in the move to increased connectivity, and eventually, automation.

The data analysed by machine learning algorithms, according to Octo, are not only transforming auto insurance, but are also perfectly placed to support the emerging concepts of connected mobility and autonomous vehicles.

The increased cyber threat to vehicle security is why telematics are so important, because it gives insurers the ability to identify who is in control of the vehicle.

It offers the ability for insurers to communicate more regularly and accurately with consumers, commented Hewett, and will encourage the type of behaviour that will improve business performance.

Telematics will play a vital role in for the future of the automotive and auto insurance industry.

This will be achieved by ensuring renewed customer satisfaction, delivering cost efficient, fair policies that prioritise safety as a driving value in the future automotive experience.

[verwijderd]
1
Following our Autopilot 2.0 report, we were told by sources familiar with the program that the Bosch Model S prototype features very similar level 3 autonomous features as the upcoming Autopilot 2.0. A Level 4 fully autonomous driving system would likely only be enabled by adding camera coverage all around the vehicle, but the Autopilot 2.0 sensor suite would serve as a starting point.

While the Bosch prototype features both radar and lidar sensors, Tesla is hoping to achieve similar coarse point cloud maps, something easier to achieve with lidar, with inputs from the radar through “temporal smoothing”.

een nieuwe aanwijzing voor Autopilot 2.0, gaat toch wel richting RoadDNA
zie ook de video van Bosch:
ze zegt dat het werkt op de snelweg in Duitsland en Californie
ook een duidelijke hint
electrek.co/2016/09/06/tesla-autopilo...
[verwijderd]
0
quote:

seren schreef op 6 september 2016 23:27:

Following our Autopilot 2.0 report, we were told by sources familiar with the program that the Bosch Model S prototype features very similar level 3 autonomous features as the upcoming Autopilot 2.0. A Level 4 fully autonomous driving system would likely only be enabled by adding camera coverage all around the vehicle, but the Autopilot 2.0 sensor suite would serve as a starting point.

While the Bosch prototype features both radar and lidar sensors, Tesla is hoping to achieve similar coarse point cloud maps, something easier to achieve with lidar, with inputs from the radar through “temporal smoothing”.

een nieuwe aanwijzing voor Autopilot 2.0, gaat toch wel richting RoadDNA
zie ook de video van Bosch:
ze zegt dat het werkt op de snelweg in Duitsland en Californie
ook een duidelijke hint
electrek.co/2016/09/06/tesla-autopilo...

@Seren, doesn't make sense to me.
--Ten eerste beschikt de Tesla vloot nú al over meer wegen waar lane-guidance mogelijk is, dan de dekking van de Tom2 HD-maps. De gewone navigatiekaarten zijn hiervoor namelijk niet geschikt en de Tesla rijdt in meer landen rond dan waar Tom2 HD-maps dekking heeft. Dat zou in mijn ogen betekenen dat Tesla op dat moment twee systemen moet gaan hanteren, danwel het éne systeem in het andere merged. Nou, dat denk ik dus niet.
--Ten tweede wordt er aangenomen dat RoadDNA in een auto zónder Lidar en ook zónder Radar kan werken. Gewoon met camera-technologie dus. Dat mag wel waar zijn, maar de laatste Tesla/Musk tweets gingen nu nét voornamelijk over resultaten die ze met Radar pointclouds) boekten. Voor RoadDNA heb je dit niet nodig, dus duidt het mijns inziens meer op een eigen aanpassing van wat ze al in huis ontwikkeld hebben, dan overstappen op iets totaal nieuws zoals RoadDNA.

De ontknoping volgt binnenkort.
Dokus
0
quote:

xynix schreef op 6 september 2016 13:13:

[...]
Waarom zou NVidia niet gewoon als een schoenmaker bij hun leest blijven. Ze hebben de komende tijd voldoende te doen in hun kerncompetentie: GPU's bakken! Even een mapmaking fabriek met wereldwijde dekking opzetten is een totaal andere ballgame. Ze kunnen wel zorgen dat ze samen met TT de sensorinfo decentraal indikken tot de relevante content om de kaarten te onderhouden.

Mobiele datacommunicatie zal toch voorlopig nog steeds een schaarse resource blijven, waar je zuinig mee om zult moeten springen.
je begrijpt het verkeerd. Wie heeft het over maps opslaan als je ze on the fly kan genereren!? HDmaps van je omgeving maken terwijl je rijdt!!!

DAT is de toekomst. TomTom, here, straks helemaal niet meer nodig. Statische data terwijl de auto het genereert waar je bij staat.

nb: heb het alleen over de HDmaps zodat de auto weet wat er om hem heen is. lane/signs/objects/position/etc
Dokus
0
quote:

Hik schreef op 6 september 2016 23:48:

[...]
@Seren, doesn't make sense to me.
--Ten eerste beschikt de Tesla vloot nú al over meer wegen waar lane-guidance mogelijk is, dan de dekking van de Tom2 HD-maps. De gewone navigatiekaarten zijn hiervoor namelijk niet geschikt en de Tesla rijdt in meer landen rond dan waar Tom2 HD-maps dekking heeft. Dat zou in mijn ogen betekenen dat Tesla op dat moment twee systemen moet gaan hanteren, danwel het éne systeem in het andere merged. Nou, dat denk ik dus niet.
--Ten tweede wordt er aangenomen dat RoadDNA in een auto zónder Lidar en ook zónder Radar kan werken. Gewoon met camera-technologie dus. Dat mag wel waar zijn, maar de laatste Tesla/Musk tweets gingen nu nét voornamelijk over resultaten die ze met Radar pointclouds) boekten. Voor RoadDNA heb je dit niet nodig, dus duidt het mijns inziens meer op een eigen aanpassing van wat ze al in huis ontwikkeld hebben, dan overstappen op iets totaal nieuws zoals RoadDNA.

De ontknoping volgt binnenkort.
volgens mij heeft tt voor roaddna al zo goed als wereld dekking. Hdmaps daarintegen nog niet
Dokus
0
quote:

Dokus_ schreef op 7 september 2016 08:22:

[...]

volgens mij heeft tt voor roaddna al zo goed als wereld dekking. Hdmaps daarintegen nog niet
kan niet meer editten; heb het even opgezocht; in laatste PB over hdmaps wordt ook vermeld dat ze voor de 17 staten roaddna hebben toegevoegd. Blijkbaar dus nog geen werelddekking.
3.904 Posts, Pagina: « 1 2 3 4 5 6 ... 93 94 95 96 97 98 99 100 101 102 103 ... 192 193 194 195 196 » | Laatste
Aantal posts per pagina:  20 50 100 | Omhoog ↑

Meedoen aan de discussie?

Word nu gratis lid of log in met uw e-mailadres en wachtwoord.

Direct naar Forum

Detail

Vertraagd 10 mrt 2025 13:45
Koers 4,706
Verschil +0,010 (+0,21%)
Hoog 4,842
Laag 4,654
Volume 65.054
Volume gemiddeld 320.944
Volume gisteren 255.240

EU stocks, real time, by Cboe Europe Ltd.; Other, Euronext & US stocks by NYSE & Cboe BZX Exchange, 15 min. delayed
#/^ Index indications calculated real time, zie disclaimer, streaming powered by: Infront