This is the place to discuss issues that are relevant for locations in the US. For any other discussions, please use the main forums.

Post Reply

"Mini onramp/offramp" -- when to model with right angles?

Post by DwarfLord
Small roads that meet large roads between major intersections -- such as parking-lot feeder roads -- often involve an angled/curved "miniature offramp" from the main road and an angled/curved "miniature onramp" returning to it. For example:

https://www.waze.com/editor/?env=usa&lo ... 1,68884045

Representing these literally using AGC constructions, as shown in the example, is extremely popular. But is that really the best solution in most cases? I'd like to understand the tradeoffs of modeling them with AGC constructions versus ordinary right-angle junctions.

The pros of using AGC constructions include:
- Advances timing of voice alert
- "Exit" voice instruction may sometimes make more sense than "turn"
- No instruction given when joining main road, so Waze less chatty
- Matches satellite image, a more literal representation of what's there

The cons include:
- Advanced voice alert may conflict with nearby turns; distance to actual road will be off
- "Turn" instruction at "offramp" will in many (most?) cases make more sense than "exit"
- Any turn instructions at "onramp" will be for NEXT junction, not this one, possibly confusing
- New editors may view as role model for matching the satellite image even at high zoom
- Display of diagonal segments may confuse drivers if real junctions are logically right-angle
- Additional map complexity, especially when only short distances involved

I would propose a more aggressive perspective that these intersection types (do they have a name?) be preferentially modeled as right-angle junctions, with the "burden of proof" resting on the the idea of making them into mini onramp/offramp AGC-type intersections. Thoughts?
DwarfLord
Wiki Master
Wiki Master
Posts: 2512
Has thanked: 1065 times
Been thanked: 1451 times

POSTER_ID:16850907

1

Send a message

Post by DwarfLord
PleaseDriveFast wrote:If GPS points show a large gap (200ft?) between the perpendicular angle and where cars are actually entering the next segment, then I think we should use the divided AGC construction as shown here. This appears to be about 100ft so it shouldn't create any map problems.
My "magic number" is 75 m (246 ft), comparable to your suggestion (I try to think in meters when using Waze as it's easy to get lengths in meters using WME).

However, there are other considerations besides distance:
- For advanced "offramp", increased potential of conflict with other nearby turns
- Potential confusion if driver can see that the actual road is some distance ahead
- New editors may see and think all such connections should be divided to match aerials
- Waze may snap drivers to the parallel "offramp" and give a bad route

After weeks of turning this over in my head from the perspective of AGCs in general, I've come to feel that this is the solution: we editors should be less concerned with the final instruction and more concerned with the penultimate instruction.

It's nice of course when the final instruction issues in time to rescue a driver who missed the penultimate instruction. But it's not a primary goal that we should sacrifice other things to accomplish.

My primary example is left turns. We editors rarely try the contortions necessary to advance turn warnings for left turns at complex intersections. Instead we assume that the driver will hear the penultimate instruction, and will interpret the final instruction as a simple confirmation once they are already positioned for the left turn. If that concept is acceptable for left turns, it should be OK for right turns as well.

If one abandons the idea of advancing the final instruction as far as necessary to rescue inattentive drivers, many AGC-like constructions become unnecessary. There are still other reasons to have them of course. I've been working on a summary of this that I might offer as an update to the AGC wiki some day.
DwarfLord
Wiki Master
Wiki Master
Posts: 2512
Has thanked: 1065 times
Been thanked: 1451 times
Send a message

Post by PleaseDriveFast
Being the last editor who touched this (I wonder what I did?), I would actually agree with the 90º AGC construction for these small lots. I would think of these as divided exit/entrances as opposed to on/off "ramp" since ramps are a specific road type.

If GPS points show a large gap (200ft?) between the perpendicular angle and where cars are actually entering the next segment, then I think we should use the divided AGC construction as shown here. This appears to be about 100ft so it shouldn't create any map problems.
PleaseDriveFast
Posts: 1262
Has thanked: 288 times
Been thanked: 645 times
Send a message