TMC is an independent, primarily volunteer organization that relies on ad revenue to cover its operating costs. Please consider whitelisting TMC on your ad blocker and becoming a Supporting Member. For more info: Support TMC
Start a Discussionhttps://teslamotorsclub.com/tmc/tags/

Lane and other road marking standardization for better autonomy/driver assist

Discussion in 'Autonomous Vehicles' started by bhzmark, May 9, 2017.

  1. bhzmark

    bhzmark Supporting Member

    Joined:
    Jul 21, 2013
    Messages:
    1,683
    A big problem with current driver assistance technology seems to be the wide variation in lane and other road markings. A lot of that could be improved with simply better lane markings and more consistent standards.

    Does anyone know of any lobbying or other efforts to try to get the various jurisdictions to implement better rules requiring better lane markings?
     
  2. BrettS

    BrettS Member

    Joined:
    Mar 28, 2017
    Messages:
    388
    Location:
    Orlando, FL
    It is an interesting thought, but honestly, there are really already a lot of standards about road markings, sign sizes, locations, and frequencies etc in most countries. Furthermore, given the billions of miles of roads and billions of signs that are already out there the chances of getting a significant number of roads repainted and/or signs replaced to meet the new standards would be fairly low. It would also take a considerable amount of money... something that many smaller jurisdictions just don't have. Also, consider that a visual system to read signs is really less than ideal... it takes a lot of processing power to scan an image to identify and read a sign. Even when that sign is a standard size in a standard location with a standard font.

    I think what would be better would be a much more detailed map system that stores data about speed limits and stop signs and other road signs so it's not necessary to automated cars to actually visually identify them. Ideally when a jurisdiction made a change and installed a new stop sign or changed a speed limit they could also submit a change to get the map system updated with the new information so that we don't need to rely on cameras to pick up that new information.

    The one infrastructure change that I think might be helpful would be for traffic lights to broadcast their state data with some sort of reasonably low power/low range RF transmitter. Again, this would mean that automated cars wouldn't need to visually identify red or green lights and ideally would also know exactly when the light will change so it may be able to speed up a little or slow down a little to be able to handle the intersection more smoothly. Again, though, this would be a very expensive and labor intensive change to update all of the traffic lights with such a system.
     
    • Like x 1
  3. oktane

    oktane Active Member

    Joined:
    Oct 25, 2016
    Messages:
    1,427
    Location:
    USA
    Trump infrastructure project should be converting carpool lanes to AP friendly lanes....whatever that entails. Magnets to mark the lanes? Make them smooth enough for 150MPH cruising....
     
  4. bhzmark

    bhzmark Supporting Member

    Joined:
    Jul 21, 2013
    Messages:
    1,683
    #4 bhzmark, May 9, 2017
    Last edited: May 9, 2017
    Changes and updates and enforcement of existing standards is what is needed. Simply requiring that more roads have lane markings, and that the lane markings be kept in good repair/visibility, could easily save lives once more AP cars are on the roads.

    It would mostly be enforcing and making changes to this: Manual on Uniform Traffic Control Devices (MUTCD) - FHWA "States must adopt the 2009 National MUTCD as their legal State standard for traffic control devices within two years from the effective date."
    2009 Edition with Revisions No. 1 and 2 Incorporated, dated May 2012 (HTML) - FHWA - MUTCD

    For instance this rule should not be "optional"

    upload_2017-5-9_18-25-42.png


    from Figure 3B-8 Long Description, Sheet 1 of 2 - MUTCD 2009 Edition - FHWA

    and this rule should be strengthened to that old markings will not be confused as current: Chapter 3A - MUTCD 2009 Edition - FHWA
    upload_2017-5-9_18-30-43.png
     
    • Like x 1
  5. RubberToe

    RubberToe Supporting the greater good

    Joined:
    Jun 28, 2012
    Messages:
    1,545
    Location:
    Pasadena, Ca
    Botts dots going away in California:

    Botts’ Dots, after a half-century, will disappear from freeways, highways

    Concerning autonomous driving: "Another major Caltrans concern: Self-driving vehicles may someday be common. Tests have shown that by covering up the striping with Botts’ Dots, the navigation systems of those vehicles can get confused, perhaps leading to drifting."

    RT
     
    • Informative x 1
  6. croman

    croman Active Member

    Joined:
    Nov 21, 2016
    Messages:
    1,536
    Location:
    Chicago, IL
    Tesla's HW2 recognizes Botts dots and it increases it's confidence. That's a poor decision by California.
     
  7. RubberToe

    RubberToe Supporting the greater good

    Joined:
    Jun 28, 2012
    Messages:
    1,545
    Location:
    Pasadena, Ca
    Good thread bhzmark. I often wondered whether any actual infrastructure improvements would be required for normal use of autonomous vehicles.

    Here is a story from over 20 years ago about a pilot test in California using magnets embedded in the road:

    IT'S NOT HARD TO GET USED TO MAGNETS DOING THE STEERING

    Given the way technology advances, you don't want to shell out an insane amount of money for something that can be easily solved some other way. A good example of this is the old style emergency phones (Call Box) along the highways. Before cell phone coverage, can you imagine the cost to string wire or connect each of these to a microwave tower in the distance?

    If I had to guess I would say that going forward the solution to any problems with autonomous driving will be solved most economically with in vehicle hardware and software. While the idea of having uniform lane markings to better enable cars to drive themselves seems almost like a no brainer, you wouldn't want to have the vehicle solely reliant on such markings. Markings can be obscured by a variety of one time events, many of which can't be anticipated (i.e. paint spills, snow, new pavement, etc.). Factor in automated machine learning about the environment, and sharing that data with other cars, and you even cover most edge cases. For example, automated car on the freeway comes upon a large dead deer in the center of the right lane. In vehicle software thats constantly "looking" for things that would normally trip it up, encounters this and informs a central database that something is amiss. The car then goes around the obstacle. The central facility can dispatch a crew to fix the problem. In the interim, cars along that stretch can be informed about the blockage prior to arriving to smooth traffic flow out. Each car would need to be able to respond as the "first arriver", but later arrivers being alerted can automatically switch to other lanes in advance, or take a different route if thats the better solution. Vehicle to vehicle, and/or vehicle to server to vehicle could all be used in some capacity.

    Given the total mileage of two lane unmarked roads in the U.S., let alone the lesser developed countries, I would think that solving the problem in software is the optimum way forward. Sure, some companies are going to perfect the approach sooner and/or better than others. But looking out say 5 years, I would think that any company that is implementing self driving for vehicles will have "solved" how to have the car drive safely on two lane unmarked roads.

    RT
     

Share This Page