0

I am working on a simulation that places radio unit at traffic signals, is there any way to generate locations of traffic signals that make the simulation realistic? Lets say we have 200 × 200 m area in a commercial block in a city. what is the best way to get Locations of N traffic signals in that area such that the distribution of traffic signals is similar to what we find in reality?

I tried to search for such a thing but apparently I am not using the right phrasing of my problem. I would really appreciate any help.

TZakrevskiy
  • 22,980
Alaa
  • 103
  • What an interesting question! I would think it depends on what kind of area we're looking at. Are you assuming we're looking downtown in a major city? – Steve Feb 08 '22 at 15:33
  • 1
    This is far too vague. Obviously, traffic signals are associated with roads and with traffic on those roads. Absent information on that network, there is nothing to be said. – lulu Feb 08 '22 at 15:34
  • @Steve Yes, a commercial block in a city – Alaa Feb 08 '22 at 15:34
  • @lulu what type of information is needed to make this question more relatable ? – Alaa Feb 08 '22 at 15:36
  • A detailed description of all the roads, along with traffic flow along them. To state the obvious: this is not a math problem (at least, not one that can be solved by reference to some standard distribution). – lulu Feb 08 '22 at 15:39
  • @lulu if I am working on a simulation that has traffic signals, is there any way to generate locations of traffic signals that make the simulation realistic ? – Alaa Feb 08 '22 at 15:41
  • 2
    Again, this is far too vague. What does "realistic" mean? One thought would be to look at data from thousands of comparable commercial blocks and model it on that (a signal goes up if peak flow, or maybe average flow, through an intersection exceeds some level....something like that). I wouldn't start with probabilistic assumptions, though the data may force you to allow uncertainty. But, to stress, if your goal is to be "realistic", then this is a data problem. – lulu Feb 08 '22 at 15:49
  • 200m is just a few blocks - that's at most 25 square blocks, so at most 25 lights (one at each intersection). Put lights at all the locations near the center and sprinkle a few along the edges. Run your simulation with several such arrangements and see what happens. Voting to close as too vague. – Ethan Bolker Feb 08 '22 at 16:15

1 Answers1

2

Neat problem. Apparently google has some capability to show stop lights on maps: https://arstechnica.com/gadgets/2020/08/google-maps-starts-widely-labeling-traffic-lights-in-the-us/. Not sure how well that works. One way to tell might be to review Google's Traffic layer during rush hour and spot intersections based on red and yellow coloring.

I'd expect the probability of there being a streetlight at an intersection, as opposed to other traffic control systems like stop signs, depends on local population density, average velocity of traffic, number of forks off the main drag, proximity to downtown, areas of frequent merging. Would also depend on districting. Streetlights are more common in commercial areas than residential areas.

There might be an opportunity to use artificial intelligence to sort out patterns.

TurlocTheRed
  • 5,683