Sidewalk Behavior at Intersections

12-19-2017 11:42 AM
Occasional Contributor III

I am trying to build a sidewalk with a parkway (landscaped) area between the sidewalk and the street. Its a fairly straight forward v split that works fine until I get to the intersections where the crosswalks lead to patch of grass.

By linking the object attribute to valency, I am able to access some of geometry and get something close to what I am looking for (not ideal).

I can offset the crosswalks the width of my  parkway for better alignment, but it only makes the crosswalks on the bottom of the image look even more disconnected.  Does anyone have any suggestions in how I can better access the sidewalk geometry to get this looking better? Like a split.index or some other property? Ideally, the concrete area in the bottom middle goes way and there is only concrete in the sections that are adjacent to the crosswalk.

Thanks for your help.

0 Kudos
5 Replies
New Contributor II

Hi Devin,  I dealt with this with a complex set of bool/switch functions. I always also add a split that adds sidewalk space before the intersections (there is a stopping-sight distance concern having trees too close to the intersection etc). Can you potentially draw out the ideal scenario so I can advise? 

Occasional Contributor III

Thanks David. I guess it would be something like this...

MVP Frequent Contributor

Maybe adding the ADA ramps will help

0 Kudos
Occasional Contributor III

Before I can add ADA ramps I need to be able to create the sidewalk....dear Liza, dear Liza

Regular Contributor

Eric Wittner‌ 
Hi Devin and Bill, (Adding Eric for Future Discussion)

This brings up a generally challenge with the corner shapes. The CS rule does not have curb ramps because the code required to do them correctly would be hundreds of lines. There are generally two types of curb ramps, diagonal and directional, and in an ideal world, they would be used based on the crosswalk settings of all legs on an approach. Intershape communication is just starting to take shape in CE, so maybe this type of interaction will be increasingly feasible. While the time I have dedicated to this problem is limited, there are a few challenges with the current start shape that made this difficult for me in the past:

1. We only have indexes and shapes generated by the shapes. This is a problem because the complexity of the start shape is an adjustable shape parameter. Do you assume 2, 4, or 6 or more shapes that you need to work with in CGA? 

2. Programming for adjustable curb radii. How do you know where to place a curb ramp when curb radii adjustments in theory adjust start location? 
3. Different start shapes (crossings) have different sidewalk shape generation. It is clear sometimes that the sidewalk valency can provide different signals of positioning based on 

4. Hard to coordinate with Crosswalk- this relates to intershape communication. 

5. Shape controls: adding a curb ramp ideally would involve either an inserted object or a pure shape translation. Figuring out which shapes need to be removed to enable that is difficult. There is not "erase" or "clip" geometry function in CGA per se. 

That said, there might be more we can do for the case Devin is looking at. I will take a look this weekend, but generally Devin I would assume a stable shape complexity and try the following: 

  1. Component splits that join the entire shape together rather than keep entire shapes. There is some functionality supporting this. This gets you a little more stable shape (in theory? not tested). 
  2. Try replacing the corner entirely with an inserted disk with some type of centroid that would match the curb radius. This is basically an approach I have thought about with curb extensions, if you can't trust the geometry, create your own. Using the new inner rect functionality will be helpful here. 
  3. Try creating shape OVER the sidewalk. I actually did this with sharrows. Sharrows in the CS street rule are basically extruded shapes that are raised inch or so above the pavement (see sharrowHeight parameter). A similar approach might be justified here. 

If any of these work let me know...I need to give this attention at some point. 

David Wasserman, AICP