Open oxidase opened 6 years ago
Am I correct to assume that this bug explains why, in this route, OSRM says to make a slight right instead of exit to? The node connecting the starting primary
highway to the link is tagged with highway=motorway_junction
, as this is described by onsite signs as a junction, even if it is not a motorway, so OSRM should treat it as such; it doesn't matter if this is a motorway or not.
Am I correct to assume that this bug explains why, in this route, OSRM says to make a slight right instead of exit to?
Yes, this is another example of the bug.
OSRM should distinguish between channelized turn lanes and off-ramps based on whether there’s a
highway=motorway_junction
node at the beginning of the link way, not based on whether the link way is amotorway_link
ortrunk_link
.This way represents an off-ramp from Westwood Northern Boulevard to Race Road in Cincinnati, Ohio. Westwood Northern Boulevard is tagged as a
highway=primary
, because this is the only grade-separated interchange along the street, the rest being at-grade; additionally, there are residential abutters all along the road. Since OSRM only recognizesmotorway
andtrunk
roads as having exits, OSRM says to turn slightly right onto the ramp – incorrectly interpreting theprimary_link
way as a channelized turn lane (a “slip road” in OSRM lingo). Fortunately, OSRM Text Instructions recovers by saying “toward Race Road”, since adestination
tag is present.It isn’t entirely clear to me how OSRM distinguishes a “slip road” from a ramp, but relying on
motorway_junction
nodes would probably result in fewer edge cases. It’s entirely possible that some motorways and trunk roads are missingmotorway_junction
nodes at grade-separated interchanges, but those cases should be fixed in OSM.(From @1ec5)