narrow bridges versus narrow doorways

88 views
Skip to first unread message

Bill Bierman

unread,
May 16, 2013, 4:21:29 PM5/16/13
to recastna...@googlegroups.com
Hello.

I have a situation where there is a narrow bridge (a 'tight rope' you might call it) that an agent is permitted to walk across.  However, a doorway with the same width would not be permitted as the walls on either side would make the space too narrow.

Is this scenario solvable with Recast+Detour?

Thanks!

Mikko Mononen

unread,
May 16, 2013, 4:25:25 PM5/16/13
to recastna...@googlegroups.com
Hi,

They should behave equally. One thing that might cause the problem is aliasing. So if there are just few voxels which define the walkable area, it is possible that depending on how the walls are aligned to the voxel grid, the door frame may become impassable. Another problem could be that the doorway has arc at top, which in which case the agent representation detect collision at the head level.


--mikko


--
 
---
You received this message because you are subscribed to the Google Groups "recastnavigation" group.
To unsubscribe from this group and stop receiving emails from it, send an email to recastnavigati...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

Bill Bierman

unread,
May 16, 2013, 4:35:11 PM5/16/13
to recastna...@googlegroups.com
They should?  Isn't that operating under the assumption that a narrow bridge must be at least as wide as the agent?  What if the agent is allowed to walk on a bridge half of the agent width?

Bill Bierman

unread,
May 16, 2013, 4:38:18 PM5/16/13
to recastna...@googlegroups.com
To clarify, the situation I described in my top post is the desired behavior.  At the moment, the narrow bridge is not being marked as traversable.

Mikko Mononen

unread,
May 16, 2013, 5:34:38 PM5/16/13
to recastna...@googlegroups.com
The handling of the narrow spaces depends a lot on the ratio between the cell size and agent radius. You usually need at least 2-3 voxels there in order to make a sliver of navmesh. I usually recommend to calculate the cell size so that it is at least twice as small as the agent radius, in that case you the most narrow passage you can traverse is about 1.5 times agent radius.


--mikko

Ben Hymers

unread,
Jun 12, 2013, 10:04:51 AM6/12/13
to recastna...@googlegroups.com
Perhaps look at rcErodeWalkableArea - it shrinks the walkable area by the agent radius, which makes perfect sense for spaces bounded by walls but not so much for spaces bounded by drops, such as your tightrope example. If anything, when using cylindrical character controllers capable of being supported by points off their centre, areas by ledges should actually be expanded, though be careful if you decide to do this as you won't be giving your agents much room for error!
Reply all
Reply to author
Forward
0 new messages