Thanks Eitan,
It did help. 

I want to combine an occupancy grid, generated by non-ROS algorithm setup,
with Costmap2D grid in some another node and later use this new grid for
path planning purpose. I agree on point that if vision sensor can also
publish point cloud then Costmap2D can take over from there. But it may be
good to have sth like this where point cloud publication is not possible.

About occupancy grid in ROS. It is in for of cell and each cell have
resolution as meter/cell. I have impression that ROS always expect this
resolution to be small. Do we deal with cases where multiple laser scans are
passing through a cell. Then how it is dealt? For example, five laser beams
pass through a cell and 2 represent obstruction while 3 do not. How this is
modelled in Costmap2D?

About path planning for rectangular robots, I am hoping to get an field A*
based implementation in place.  I hope it will suffice.  I will also have a
look at SBPL planner. Has someone tried using CHOMP? 

About move_base having one node only. Am I correct if I say CostMap
generation using CostMap2DROS, global and local path planning are happening
on this move_base node? If yes, Isn't it is loaded. One would probably like
to take out say path planning as different node to other processor ! Or
Advantage of having all of them in one place is "there are no network
delays" ?

thank you
nitin




-- 
View this message in context: http://ros-users.122217.n3.nabble.com/costmap-grids-tp1051168p1060179.html
Sent from the ROS-Users mailing list archive at Nabble.com.

------------------------------------------------------------------------------
This SF.net email is sponsored by 

Make an app they can't live without
Enter the BlackBerry Developer Challenge
http://p.sf.net/sfu/RIM-dev2dev 
_______________________________________________
ros-users mailing list
ros-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ros-users