May as well say it... Line standards are needed

[Driver Dave]

This isn't a Tesla issue.

Many cars are getting line following driver assistance. Basically cruise control plus.

Many cars soon will become all cars.

Currently lines are not well maintained, use paint that fades quick, and have varying standards across towns and states.

There needs to be a nationwide standard for line painting that has the same standards for what is what, uses paint that lasts longer, with rules about how they are maintained.

Its sort of obvious that "line following driver assistance" will soon become adopted on a mass market level, and that plans for having the best lines to support them need to get started.

 

[Derek Kessler]

Agreed. I'm happy to see that, at least in Ohio, whenever they resurface or repaint a highway they're putting down dotted lines to mark merging and diverging lanes. It gives lane guidance cameras something to watch instead of drifting to the right as the perceived lane widens.

 

[croman]

NHSTA has standards for interstates and US highways. Local municipalities have the freedom to do what they want and I don't see that easily changed in the US.

 

[jaguar36]

Yeah, we really need a real standard, lets make a new one!

 

[ItsNotAboutTheMoney]

Yeah, we really need a real standard, lets make a new one!

One for every occasion.

Realistically, the systems need to deal with weather, dirt, crud, and wear. A line standard is a good thing, but we simply can't depend on it.

 

[alcibiades]

This isn't a Tesla issue.

Many cars are getting line following driver assistance. Basically cruise control plus.

Many cars soon will become all cars.

Currently lines are not well maintained, use paint that fades quick, and have varying standards across towns and states.

There needs to be a nationwide standard for line painting that has the same standards for what is what, uses paint that lasts longer, with rules about how they are maintained.

Its sort of obvious that "line following driver assistance" will soon become adopted on a mass market level, and that plans for having the best lines to support them need to get started.

Line consistency is neither necessary nor sufficient for us to move past AP as is. If the goal is simply that cars follow lines better, huzzah. But that doesn't matter much, since the exact same driver-awareness will be necessary given the limitations of the system's orientation to the world. You are very clear in the 101 threads that yo believe it's the driver's fault in edge cases. That will be no different in your proposal.

We should have L2 autonomy systems that use mapping. Even the best line consistency in the world isn't immune to bad weather or a liquid spill or eventual road wear of super-paint or real-world state/fed budgets. So while there might be fewer edge cases with magic lines, it's a not a paradigm shifter for the necessity of the driver always to be paying attention.

A vision system that can identify static objects or can identify things that are marked with warning or construction indicators would be more useful than line consistency. Is this in principle capable from our current "vision" system? Maybe. Everyone else says lidar is needed. Even if lines are painted super-duperly, the inability of a vision system to discern objects leave Tesla AP as requiring massive vigilance.

So sure, let's spend gazoodles fixing lines. It'll help drivers who don't have lane-assist, which is basically best for all of us. Better lines should be termed as "assistance to autopilot" but autopilot can't rely on there always being good lines, so autopilot also needs driver assistance at all times.

 

[Driver Dave]

Line consistency is neither necessary nor sufficient for us to move past AP as is. If the goal is simply that cars follow lines better, huzzah. But that doesn't matter much, since the exact same driver-awareness will be necessary given the limitations of the system's orientation to the world. You are very clear in the 101 threads that yo believe it's the driver's fault in edge cases. That will be no different in your proposal.

We should have L2 autonomy systems that use mapping. Even the best line consistency in the world isn't immune to bad weather or a liquid spill or eventual road wear of super-paint or real-world state/fed budgets. So while there might be fewer edge cases with magic lines, it's a not a paradigm shifter for the necessity of the driver always to be paying attention.

A vision system that can identify static objects or can identify things that are marked with warning or construction indicators would be more useful than line consistency. Is this in principle capable from our current "vision" system? Maybe. Everyone else says lidar is needed. Even if lines are painted super-duperly, the inability of a vision system to discern objects leave Tesla AP as requiring massive vigilance.

So sure, let's spend gazoodles fixing lines. It'll help drivers who don't have lane-assist, which is basically best for all of us. Better lines should be termed as "assistance to autopilot" but autopilot can't rely on there always being good lines, so autopilot also needs driver assistance at all times.

I do agree that better tech and better driver ability (knowing how to drive with AP) is needed.

But... I worry less about Tesla drivers and more about all the other manufacturers who won't put in the high tech state of the art systems, but will just slap in a mobile eye chip in every car and start the marketing.

So from that point of view, I think it is fair to say "we should start to plan for better, or just enforced, standards of roads and lines"

 

[Driver Dave]

Looks like it is not just line following tech, but line following humans as well.

San Jose: Fatal Greyhound bus crash blamed on poor road markings

 

[SMAlset]

I'm waiting for California's Caltrans to say they don't have the budget/manpower to update the road markings and signage so there's not much they can do. Hard enough to get some roadway potholes and cracked sinking payment areas repaired as it is in a timely fashion. Don't now about other States, but our transporation budget has been raided numerous times over the years to pay for other things so I can completely see this reasoning being given.

Tesla and other manufacturers offering vehicles with driver assist features know how bad the roadways are across the U.S. and know that right now we have to have driver's in control over their driving. I don't see that changing any time soon.

 

[EV-lutioin]

I'm waiting for California's Caltrans to say they don't have the budget/manpower to update the road markings and signage so there's not much they can do. Hard enough to get some roadway potholes and cracked sinking payment areas repaired as it is in a timely fashion. Don't now about other States, but our transporation budget has been raided numerous times over the years to pay for other things so I can completely see this reasoning being given.

Tesla and other manufacturers offering vehicles with driver assist features know how bad the roadways are across the U.S. and know that right now we have to have driver's in control over their driving. I don't see that changing any time soon.

Keep in mind that highway maintenance funds in CA come partly from gas taxes and EVs don't use gas. That is why I am in favor of Governor Brown's gas tax increase and EV surcharge fee, despite extra costs for me (2 EVs). At this point, it appears that CalTrans can't even afford paint. $25M of the funds annually go to Freeway Service Patrols, which would help with maintaining lines.

 

[SMAlset]

Keep in mind that highway maintenance funds in CA come partly from gas taxes and EVs don't use gas. That is why I am in favor of Governor Brown's gas tax increase and EV surcharge fee, despite extra costs for me (2 EVs). At this point, it appears that CalTrans can't even afford paint. $25M of the funds annually go to Freeway Service Patrols, which would help with maintaining lines.

Yes, we are aware (and why voters got smarter approving tax hikes only tied to bills restricting the money's use for what was intended by the voters). We ourselves don't have any problems as EV owners to share in the cost of road repair/maintenance. I haven't been paying attention to how they propose to make it fair for all (ICE and EVs). Levying one flat rate for EVs that don't drive alot wouldn't be fair to them however. For ICE vehicles how little you drive would be factored into how much gas you have to buy. Don't want to get off the topic of line standards here.

 

[Economite]

Keep in mind that highway maintenance funds in CA come partly from gas taxes and EVs don't use gas. That is why I am in favor of Governor Brown's gas tax increase and EV surcharge fee, despite extra costs for me (2 EVs). At this point, it appears that CalTrans can't even afford paint. $25M of the funds annually go to Freeway Service Patrols, which would help with maintaining lines.

Freeway Service Patrols are tow trucks that look for stranded cars (usually in the shoulder, but sometimes in lanes) and move them off the freeway. Those folks don't (and really can't) maintain lane markings.

Repainting lane lines on a freeway is a pretty major operation. It's a much bigger job than repainting the lines on a parking lot.

The paint isn't really paint, but a thick, cement-like goo. And old paint has to be scrubbed off before the new paint is applied. The work is done at night because it involves closing off lanes to traffic. This all takes a multi-person crew with several trucks. It probably should be done more often than it is done in CA, but there's little chance that it would be possible to always have the lines perfect (or even almost perfect). Lines just wear out, and not necessarily evenly.

Ultimately, autonomous vehicles will need to deal properly with roads as they exist. An AV that depends on perfect or uniform roads will be giving up (ie requiring driver intervention) or crashing way too often.

 

[Economite]

Repainting lane lines on a freeway is a pretty major operation. It's a much bigger job than repainting the lines on a parking lot.

Found this interesting article: Stripes on Interstate 5 to pave way for self-driving vehicles

If I'm reading this correctly, a project to restripe a roughly 200-mile long basically rural (basically two lanes in each direction) stretch of I-5 in is expect to cost Caltrans $6.8 Million. That's about $34,000 per mile, and the cost would be much higher, presumably, if this was an urban freeway with more lanes. And the work (even with Caltrans standards) needs to be redone every two years.

 

[EV-lutioin]

Found this interesting article: Stripes on Interstate 5 to pave way for self-driving vehicles

If I'm reading this correctly, a project to restripe a roughly 200-mile long basically rural (basically two lanes in each direction) stretch of I-5 in is expect to cost Caltrans $6.8 Million. That's about $34,000 per mile, and the cost would be much higher, presumably, if this was an urban freeway with more lanes. And the work (even with Caltrans standards) needs to be redone every two years.

Yes! and right down the road from me. Way to get into the game CalTrans!

 

[Economite]

Way to get into the game CalTrans!

Each of the twelve geographic CalTrans districts operates fairly independently; so more like "Way to get into the game CalTrans District 2." I have a feeling that this might be more of a priority in the rural districts, where it is relatively common at night for the only lighting on the road to be the headlights of your own car. Those highways can feel pitch black at night, and really need the extra reflectivity.