How will autonomous cars impact highway design?

[kernals12]

In a future where cars can communicate and react accordingly, how will highway design change?

I imagine that left hand ramps would become popular as they require less right of way and are cheaper since they would not be flustered merging into the fast lane. Also, with their precise coordination, they would require shorter merge and weave segments.

[hotdogPi]

How does a left-hand ramp use up less space than a simple diamond?

[kernals12]

How does a left-hand ramp use up less space than a simple diamond?

Because regular diamonds need a certain distance between the two intersections. SPUIs and TUDIs fix this, but require wider bridge structures that cost more

[kphoger]

But each one has different capacities, so you can't necessary compare them directly like that.

[jeffandnicole]

In a future where cars can communicate and react accordingly, how will highway design change?

I imagine that left hand ramps would become popular as they require less right of way and are cheaper since they would not be flustered merging into the fast lane. Also, with their precise coordination, they would require shorter merge and weave segments.

But, if autonomous vehicles became the standard, as you have previously mentioned highway capacity would be increased as vehicles can drive closer to each other, so there would be no reason to reconstruct what already exists.  We aren't going to see many new highways being built, and with a high percentage of autonomous vehicles being a long ways off, we won't see left sided ramps be the norm for quite some time...if ever.

Also, highway design doesn't always mean the cheapest option wins.  Concrete jersey barriers are much more expensive than guardrail, which is more expensive than cable, which is more expensive than nothing.  But, safety considerations warrant spending that money.  12 foot lanes cost more than 10 foot lanes.  Traffic lights cost more than stop signs.  Traffic lights with vehicle detection and left turn arrows cost more than a fixed timed RYG light.  Cost is a considerable factor, but by far not the only one.

[Scott5114]

I don't think highway design will change all that much due to autonomous cars. Manual control will always be there as a fallback, so you don't want to introduce features that a human would have problems handling. I also think in America, particularly in certain states, there would be political opposition to making freeways automatic-only. What's the point in owning a classic car if you're not allowed to drive it?

I think if there are changes to highway design, it will be subtle things that make the roads more machine-readable, in the way that UPC barcodes replaced price stickers. You might see lane striping go from being broken lines to a solid alternating white-black or something like that. Or white may be replaced by a color that is easier to distinguish from the pavement, like fluorescent yellow-green.

[Joe The Dragon]

I think if there are changes to highway design, it will be subtle things that make the roads more machine-readable, in the way that UPC barcodes replaced price stickers. You might see lane striping go from being broken lines to a solid alternating white-black or something like that. Or white may be replaced by a color that is easier to distinguish from the pavement, like fluorescent yellow-green.

So auto cars can be hopeless in work zones?

[Rick Powell]

Theoretically, with all cars being autonomous, the forgiving ramp lengths and angles, curve radii, stopping sight distance and other elements could be tightened far more than they are currently because many of these elements take driver error and behavior into account. I suppose there is machine error too, but it will likely be shown to be way more predictable, leading to measures only being placed in critical places.

[Scott5114]

I think if there are changes to highway design, it will be subtle things that make the roads more machine-readable, in the way that UPC barcodes replaced price stickers. You might see lane striping go from being broken lines to a solid alternating white-black or something like that. Or white may be replaced by a color that is easier to distinguish from the pavement, like fluorescent yellow-green.

So auto cars can be hopeless in work zones?

I imagine some DOT (probably Oklahoma's) would make the mistake of not designing their workzone to be machine-readable...once. The other 49 states would blast out change orders to their work zones after seeing something that looked like I-35W in Fort Worth on the national news.

[kphoger]

Manual control will always be there as a fallback, so you don't want to introduce features that a human would have problems handling.

Quoted for emphasis.   

[kernals12]

In a future where cars can communicate and react accordingly, how will highway design change?

I imagine that left hand ramps would become popular as they require less right of way and are cheaper since they would not be flustered merging into the fast lane. Also, with their precise coordination, they would require shorter merge and weave segments.

But, if autonomous vehicles became the standard, as you have previously mentioned highway capacity would be increased as vehicles can drive closer to each other, so there would be no reason to reconstruct what already exists.  We aren't going to see many new highways being built, and with a high percentage of autonomous vehicles being a long ways off, we won't see left sided ramps be the norm for quite some time...if ever.

Also, highway design doesn't always mean the cheapest option wins.  Concrete jersey barriers are much more expensive than guardrail, which is more expensive than cable, which is more expensive than nothing.  But, safety considerations warrant spending that money.  12 foot lanes cost more than 10 foot lanes.  Traffic lights cost more than stop signs.  Traffic lights with vehicle detection and left turn arrows cost more than a fixed timed RYG light.  Cost is a considerable factor, but by far not the only one.

It could be as simple as providing you with an electronic alert that a car is merging from a left sided ramp. Or a more high tech solution would be to have the two cars cooperate their speeds to ensure a smooth merge.

[GaryV]

Gotta make the cars autonomous first.

There was another Tesla crash this week in Lansing, MI.  Using "Autopilot", the car crashed into a police car with its lights flashing.

[vdeane]

I imagine the first change, especially in the near term, is the increased emphasis on uniform nation-wide standards, which we're already seeing.  I believe the 2020 MUTCD draft specifically calls out autonomous cars as a reason for this.

[jeffandnicole]

Gotta make the cars autonomous first.

There was another Tesla crash this week in Lansing, MI.  Using "Autopilot", the car crashed into a police car with its lights flashing.

You left out that a driver in a non-autonomous vehicle crashed into a deer first.  So both ways of driving aren't fool-proof.  And Tesla points out that you're still supposed to pay attention and assist when needed.