REGISTER | LOGIN
Breaking News
Blog

Automakers Sharing Data? When Cars Fly

NO RATINGS
Page 1 / 2 Next >
View Comments: Newest First | Oldest First | Threaded View
imispgh
User Rank
Freelancer
NHTSA needs to step up
imispgh   4/24/2017 7:10:33 PM
NO RATINGS
This is where NHTSA is not doing its due diligence in the US. It is wasting time and needlessly putting lives at risk.  There needs to be a common set of data and standards so everyone does the minimal things, does them the same way and has the same data ground truth at all times. There needs to be a Scenario Matrix that delineates the testable minimum capability standards for each autonomous level.It needs to include all  base scenarios, variations on them, the method they should be handled and a verification that sensors data is built from different sensor types to ensure the highest accuracy possible and that it is double if not triple verified.

Map data also has to be the same. You cannot have one company make an update the other does not yet have. You cannot have some vehicles know there is an important change to the road and others not know it's there. Just like you cannot have some cars handling some scenarios and not others and handling them differently vendors to vendor.  This is where the government has value. None of the commercial players is gonna be able to drive this. There is way too much competition. distrust and petty nonsense.  If NHTSA creates a level and competent playing field it will actually ensure even competition and this technology will occur much faster and harm far less people getting there.

On that point. NHTSA has got to stop this machine learning  use of human Guinea pigs.  It is wasting time and is needlessly putting people at risk.

There is a far better way to so all of this. ML is great but limited and being used as a crutch. A crutch with significant weaknesses.

More details here

Due Diligence Recommendations for the Mobile, Autonomous and Driverless Industry

https://www.linkedin.com/pulse/due-diligence-recommendations-mobile-autonomous-industry-dekort

Les_Slater
User Rank
Author
Re: Driving policy must be an industry standard
Les_Slater   1/17/2017 9:26:43 AM
NO RATINGS
Cars will be capable of driving autonomously long before all the standards and regulations are sorted out. The onboard syestems, which may be fully autonomous, likely will not be aware of restrictions, never mind comply with them. Where prohibited, drivers and owners would be responsible.

Doug_S
User Rank
Author
Re: Driving policy must be an industry standard
Doug_S   1/16/2017 1:25:28 PM
NO RATINGS
Why would the OWNER be fined or jailed? It should be up to the car company that writes/maintains the software to insure it can only operate autonomously in areas where it is legal to do so. In areas where the legal status isn't known for sure, the vehicle should refuse to operate autonomously.

Perhaps it might include some method to override that in an emergency (i.e. you are injured and need to get to a hospital) but only in cases where the owner overrides the vehicle's restrictions should the owner be liable.

Making the owner criminally liable for what an autonomous vehicle does is the fastest way to insure they never take off. That would be like putting you in jail because you bought some consumer product that unbeknownst to you leaked RF on restricted bands.

Les_Slater
User Rank
Author
Driving policy must be an industry standard
Les_Slater   1/16/2017 11:59:01 AM
NO RATINGS
There is a regulatory issue also, whatever autonomy these vehicles will have will be very dependent on federal, state and local policy. A vehicle that may be quite capable of driving itself might find its owner fined, or even jailed if it excercizes that capability without the consent and regulatory requirements of the jurisdiction.

Not only will industry standards be necessary, but will have to be approved by government agencies. It would make sense for at least part of the infrastructure to be publically built and operations publically funded.

It seems to me that crowdsourcing road data might have advantages, but disadvantages too. Stationary cameras and other roadside sensors could be more robust. In either case though, the robustness and timeliness would be of the utmost importance.

These vehicles could be totally autonomous, not requiring a local infrastructure separate from the vehicles themselves. It would take making these vehicles part of a mesh network. Of course, here (no capital H), would definitely require industry standards.

There is a regulatory issue also, whatever autonomy these vehicles will have will be very dependent on federal, state and local policy. A vehicle that may be quite capable of driving itself might find its owner fined, or even jailed if it excercizes that capability without the consent and regulatory requirements of the jurisdiction.

Not only will industry standards be necessary, but will have to be approved by government agencies. It would make sense for at least part of the infrastructure to be publically built and operations publically funded.

It seems to me that crowdsourcing road data might have advantages, but disadvantages too. Stationary cameras and other roadside sensors could be more robust. In either case though, the robustness and timeliness would be of the utmost importance.

These vehicles could be totally autonomous, not requiring a local infrastructure separate from the vehicles themselves. I would take making these vehicles part of a mesh network. Of course, here (no capital H), would definitely require industry standards.

Victor.Lew
User Rank
Rookie
A primer to explain the background would enhance
Victor.Lew   1/16/2017 11:49:55 AM
NO RATINGS
Regarding "Automakers Sharing Data", I prefer to first be presented a primer, as a backgrounder, of the extensive efforts of the mapping, navigation, and automotive industry efforts relative to any culmination of a substantial and possibily monolthic navigation and mapping data sharing model presented in this article.

Years of major efforts to bring together many parties, from past efforts of the AdaptIVe consortium, ERTICO-ITS Europe, Vehicle and Road Automation (VRA) Group, SENSORIS Innovation Platform, Navigation Data Standard (NDS) Association, Open Autodrive Consortia, ADASIS Forum, up to the present efforts from entities such as OpenStreetMap, Google Maps, or Dynamic Map Development Co. (SIP-adus in Japan) or HERE or TomTom, would provide evidence that any outsider not directly involved in these discussions and negotiations simply cannot project any reasonable assessment of the depth, details or timing of such monolithic agreements.

Observing that even as we speak the telecommunications industry, the cloud providers, the big data analytics leadership, and other supporting providers are signing contracts for committing substantial resources to the infrastructure required to faciltate this monolithic model, even before the final announcement is consummated, supports an analysis that needs much elaboration and underpinning.

An announcement of a master data sharing plan for mapping could easily be presented to the marketplace tomorrow, and this article is then essentially void of any real analysis or accuracy, or merits.  

I am miffed why those who reside in California seem to be touted as the leading experts in the Auto Industry.

Most Recent Comments
Like Us on Facebook
EE Times on Twitter
EE Times Twitter Feed