2 Questions: "push" agents and dynamic costs for areas
57 views
Skip to first unread message
behelf...@gmx.net
May 9, 2013, 2:33:16 PM5/9/13
to recastna...@googlegroups.com
Hi there,
i managed to integrate recast into the cocos3d framework and most features now work. So now i went back evaluating possibilities and strategies. There are two major points i`m not sure how to approach.
1. Somewhere Mikko mentioned that combining recast with a physics engine to influence agents position might not be a good idea (error prone, e.g. how to keep the agent on the mesh, difficulties with how to decide who has two control the agent). So i am searching for a way how i could simulate things like an agent beeing pushed back by a force, like a shockwave or the like? My first idea was to simply set the velocity of the agent, and afterwards replan the path. Could that be a possible solution? Any other approaches i could investigate?
2. Mikko stated that using Detour/Recast as base for a influence map wouldn`t be a good idea cause they would need a denser grid than the bigger polygons could provide. Is this still true, also for small inhouse scenarios? My first thought was that a TileMesh would be a good start for a influence map?!
3. Let`s say i would have a influence map. Is there a possibility to set the area costs only for calculation? My idea was that i would make more dangerous areas more costly so that the pathfinding tries to avoid them. But of course these costs should only be considered during pathfinding and should not affect the speed of the agent. Is there any chance to make this work? Is dynamic chaning of area costs even possible without recreating the navmesh? I think of something like mapflooding to calculate the dangerous areas depending on the agents (enemy agents) positions.
4. If above things would somehow work.. how could i possibly have different area costs for different agents (the two teams)? Obvious solution would be two have two navmeshes, but i can`t wrap my head around how i could have two meshes but still having all agents consider each other while moving..
As you probably noticed i am kind of new to this stuff but so grateful for your work,
thanks,
alex
i managed to integrate recast into the cocos3d framework and most features now work. So now i went back evaluating possibilities and strategies. There are two major points i`m not sure how to approach.
1. Somewhere Mikko mentioned that combining recast with a physics engine to influence agents position might not be a good idea (error prone, e.g. how to keep the agent on the mesh, difficulties with how to decide who has two control the agent). So i am searching for a way how i could simulate things like an agent beeing pushed back by a force, like a shockwave or the like? My first idea was to simply set the velocity of the agent, and afterwards replan the path. Could that be a possible solution? Any other approaches i could investigate?
2. Mikko stated that using Detour/Recast as base for a influence map wouldn`t be a good idea cause they would need a denser grid than the bigger polygons could provide. Is this still true, also for small inhouse scenarios? My first thought was that a TileMesh would be a good start for a influence map?!
3. Let`s say i would have a influence map. Is there a possibility to set the area costs only for calculation? My idea was that i would make more dangerous areas more costly so that the pathfinding tries to avoid them. But of course these costs should only be considered during pathfinding and should not affect the speed of the agent. Is there any chance to make this work? Is dynamic chaning of area costs even possible without recreating the navmesh? I think of something like mapflooding to calculate the dangerous areas depending on the agents (enemy agents) positions.
4. If above things would somehow work.. how could i possibly have different area costs for different agents (the two teams)? Obvious solution would be two have two navmeshes, but i can`t wrap my head around how i could have two meshes but still having all agents consider each other while moving..
As you probably noticed i am kind of new to this stuff but so grateful for your work,
thanks,
alex
Reply all
Reply to author
Forward
0 new messages