Open AntonM030481 opened 2 years ago
Most likely it's too difficult to solve this problem. But a lot of small changes together or separately can be done:
The entrance is not usually your main target as a driver. You need parking.
For sure I usually need parking place. But in most of cases info about available parking options is incomplete. It's really difficult to identify ALL available parking options. This is great task but it's it's very difficult to make results not misleading.
If I select some building as my target in 99% I will need to go inside. It does not matter if my primary routing method is by foot, car, or public transport.
Naive approach is to go as close as possible to the center by primary routing method. But if we know that there are specific entrances for this building, why to ignore them?
The basic use case is: User select building as target. There are 2 roads close to it. 1st road in front of building (close to the entrance). 2nd one from back side (no entrance there). User should be routed to the 1st one. Last mile from 2nd road can be really long (sometimes it's not possible just walk around the target building).
Sometimes user is routed to the wrong side of target building (this applies to pedestrian routing too now). There are 3.000.000 entrances in OSM which can help to solve this problem.
Currently in Car navigation you are routed to the closest point (P) to the center of target object (C). And then there is a "virtual jump" from P to C. This jump is limited to not pass through any obstacle.
There are some large objects (e.g. supermarkets) which are marked with entrances or have pedestrian road right to it border (virtual entrance). It will be great to guide user to travel by car and finish his route by foot.
E.g. 53.9059044, 27.4402135