search

conveyal / r5

Developed to power Conveyal's web-based interface for scenario planning and land-use/transport accessibility analysis, R5 is our routing engine for multimodal (transit/bike/walk/car) networks with a particular focus on public transit
https://conveyal.com/learn
MIT License
289 stars 74 forks source link

Check speed for linking water to roads #337

Open abyrd opened 7 years ago

abyrd commented 7 years ago

It seems like we're still linking faraway cells to roads with minimal or no cost. This should be changed with a shorter linking distance and lower speed to traverse any >1 cell links.

abyrd commented 6 years ago

We appear to still have this problem wiith driving searches. There are still big horizontal jumps in isochrones at the coastline (see image) which imply zero travel time. image

abyrd commented 6 years ago

This might just be an illusion caused by the car being relatively fast. Still should be checked in the code to make sure speed is being properly applied.

ansoncfit commented 6 years ago

We are applying a cost: https://github.com/conveyal/r5/blob/dev/src/main/java/com/conveyal/r5/streets/LinkedPointSet.java#L320

For cars, the assumed speed is 11 m/s (https://github.com/conveyal/r5/blob/dev/src/main/java/com/conveyal/r5/profile/ProfileRequest.java#L66), which takes one pretty far into the water in 60 seconds. 🌊

Other than linking pointset points to streets, it looks like the only other place carSpeed might be used is the StreetRouter's A* heuristic. So we could consider reducing it to something closer to real driveway/parking lot speeds.

ansoncfit commented 6 years ago

:tada: This issue has been resolved in version 4.2.0 :tada:

Your semantic-release bot :package::rocket:

ansoncfit commented 5 years ago

We discussed an example in Seattle where isochrones appeared to expand into water as quickly as they did along streets with transit service. Is the correct speed being applied (OFF_STREET_SPEED_MILLIMETERS_PER_SECOND)?