Sorry. I used the search function for The Guardian and I read back to Friday in this thread. Obviously it is much older news than that but I tried.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Autonomous Car Progress
- Thread starter stopcrazypp
- Start date
-
- Tags
- Autonomous Vehicles
diplomat33
Average guy who loves autonomous vehicles
Mobileye update from their Q2 report:
- We continued to expand on our Basic ADAS leadership position including a design win with a key customer that extended our relationship through 2035 (including meaningful Cloud-Enhanced ADAS volumes). Additionally, we leveraged our computer vision expertise and massive video database to achieve European certification of the first vision-only solution for use by OEMs to efficiently comply with the European General Safety Regulation (GSR) for Intelligent Speed Assist.
- Our first major SuperVision design win beyond Geely Automotive Group was announced by Porsche during Q2 and we see excellent potential for similarly timed launches across other VW Group brands. Additionally, VW of America announced a testing program of Mobileye Drive-equipped VW ID.Buzz vehicles in Austin, Texas after only months of development and integration activity.
- Near-term SuperVision launches remain on-track and we expect to have SuperVision equipped on 5 production vehicle models (including 2 models sold outside of China) by the first quarter of 2024 as compared to 1 vehicle model at the beginning of 2023. Additionally, the software roll-out of Navigate-on-Pilot functions for Zeekr vehicles is proceeding well and garnering accolades from key China media influencers, particularly for its assertive, efficient driving style.
diplomat33
Average guy who loves autonomous vehicles
Cruise announced Nashville as their next ride-hailing city. Cruise will be in 7 cities in a year with more after that.
Also, read the whole thread. Kyle lays out reasons why he believes Cruise's software is generalized to work in any city.
Also, read the whole thread. Kyle lays out reasons why he believes Cruise's software is generalized to work in any city.
Trucking also has a few non-trivial obstacles that cars do not.
Road flares: It's impossible to deploy them as legally required without a human driver. Changing this requirement means lobbying every state they might drive through. Where long-distance trips are the most profitable, this also means going through a large number of states.
Weigh stations and inspections: How does a truck submit paperwork or follow directions on going to the scale or correct waiting spot?
Air-craft quality fly-by-wire network: Cars do not require redundant internal drive-by-wire networks but production trucks do for automation. This is a lot of effort on a new generation of hardware for multiple network interfaces which doesn't actually advance the difficult pieces of self-driving vehicles. Better to do it as a background task as new hardware is required for other reasons (higher sensitivity or durability for example).
Finally, which trucks are most useful for long highway runs they would need to fully navigate cities and could probably only service cities with a Waymo taxi service (staff to handle problems, mapping completed, etc.).
There were a few ways to pivot around those issues that they didn't take. Yard automation: lots of trucks do nothing more than move containers between crane stations for loading into stacks, ships, train cars, etc. Robotic systems exist for this but completely replace human operations because they have limited safety capabilities to detect humans for example.
Another would be replacing the 2nd driver in team trucking. If they could implement eye's off truck driving which counted as down-time for a single driver, then a single human could drive 24/7 while handling edge cases like weigh stations and flares.
Last edited:
KArnold
Active Member
Aurora is planning on driverless trucks Dallas/Houston in 2024. They do it with safety drivers today. Unlike some companies, they actually achieve what they say.
Do you know how? Chase vehicle? Multi-truck trains: single human for multiple trucks operating as a single unit? Did they manage to get an exemption to those laws?
I'm not sure how you do driverless trucks in USA on public highways without a human nearby, and look forward to seeing how Aurora worked around the regulatory restrictions.
diplomat33
Average guy who loves autonomous vehicles
I believe Aurora will have remote assistance for the driverless trucks. But just like with robotaxis, remote assistance is only for non-safety issues and only provides hints, it does not remote control the truck. So they still need the driverless trucks to be reliable enough to handle all safety issues on their own. This is especially true for trucks since they go at high speeds. Aurora has added the capability to autonomously pull over in case of an emergency and to be able to pull back onto the highway and resume driving once the issue has been solved remotely. Now you might ask about situations where the truck breaks down and can't drive anymore. Then I guess you would have a truck stuck on the side of the highway and they would need to send someone to get the truck. Obviously, they need to make sure the hardware and software is reliable enough that those instances are extremely rare. But that is one reason why it takes a long time to test and validate because you need to thoroughly test both the hardware and software to make sure that it is reliable.
That's just it. A truck pulling over on the highway without immediately (IIRC within 5 minutes) deploying flares/cones/lights is illegal in every state AFAIK. I'd like to know how they manage to handle this failure mode without having a human already very close to the truck, or receiving a waiver from following that law. Sending someone out to pick it up in the event of a failure doesn't work for trucks.
Of course, I suppose they could opt to simply ignore a few laws and ask for forgiveness when someone notices. Personally, I despise that approach to developing tech.
Last edited:
Knightshade
Well-Known Member
It's within 10 minutes- in the US anyway- but otherwise correct.
49 CFR § 392.22
(b) Placement of warning devices—(1) General rule. Except as provided in paragraph (b)(2) of this section, whenever a commercial motor vehicle is stopped upon the traveled portion or the shoulder of a highway for any cause other than necessary traffic stops, the driver shall, as soon as possible, but in any event within 10 minutes, place the warning devices required by § 393.95 of this subchapter, in the following manner:
As to how they will place the warning signs or flares, that's easy--- Each truck can carry a Teslabot to deploy them
diplomat33
Average guy who loves autonomous vehicles
Well, I think the obvious solution would be to have a human in the passenger seat of the driverless truck. So it would still be driverless since there is no human in the driver seat. But there would a human riding as a passenger that could deploy flares and cones in the unlikely event the truck has to pull over.
Knightshade
Well-Known Member
Unclear what your gain in bothering with the trouble and expense to automate the truck is then? I mean I guess you save the slightly higher pay to someone with a CDL that would eventually offset the cost of the tech?.... and MAYBE you reduce accident rates by some amount?
Drivers are limited in the number of hours they can log in a day. Driverless could run 24/7 without stopping (except for refueling). The passenger in the truck wouldn't have to be alert to much so could sleep all the time if they wanted. It would be an incredibly shitty job but it would save some money.
Of course, in my scenario trucks are running 24/7 so inspections would have to be much more frequent. Which would introduce delays and costs.
It will be interesting to see where the cost/benefit eventually works out to be.
KArnold
Active Member
Both Waymo and Aurora were applying for a waiver to those rules:
Trucker unions are of course opposed to it:
/cloudfront-us-east-2.images.arcpublishing.com/reuters/SL2TPYZOHFNYJBBFVJ5FTHSLQI.jpg)
US union opposes driverless trucks waiver for Waymo, Aurora
A major U.S. transport union on Tuesday opposed a request by Alphabet Inc's self-driving unit Waymo and autonomous driving technology company Aurora Innovation Inc for an exemption from rules on warning devices for large semi-trucks, citing safety issues.
www.reuters.com
Met some of the key people on aurora - very cool and they are doing logistics centers to logistics centers. Smart
Who would be responsible for the security of the load? A Chain Binder comes loose a Tie Down breaks. The Load shifts. Would it be difficult to train every Truck to back into any loading dock to pick up and drop off any Trailer? If the Semi is just doing the exact same route and using the same types of Trailers all the time self driving Semis would most likely work. I don't see it working very well if you have to transport different loads on different kinds of Trailers. One day you may be delivering groceries in a Box Trailer. The next day you are delivering a Caterpillar on a Low Boy out to the Country.
Exactly. They are going to be most useful running from logistic center to logistic center. UPS, FEDEX, some of the major freight carriers. Small companies have way to much diversity in loads and those loads can end abruptly.
KArnold
Active Member
I believe the autonomous vendor would be responsible for the load.
Load shifts are detectable. See the above case study where a vehicle hit the side of a trailer. It was detected, identified, and the truck pulled over.
With Aurora (and probably others) the autonomous vehicle pulls into a staging area. Manual or automated mules move it to/from the loading docks. At least for now.
I'm sure you are right there will remain many instances where self driving trucks are not appropriate. But for most long hauls over interstates using closed box trailers it is an incredible benefit - potentially.
Similar threads
- Article
