[ros-users] tf with multiple localisers

Bhaskara Marthi bhaskara at willowgarage.com
Wed Oct 27 16:11:46 UTC 2010


On Tue, Oct 26, 2010 at 11:10 PM, nickopen <nicholas.hillier at gmail.com>wrote:

>
> Apologies if this is a repost - I had a search and a bit of a read but
> couldn't easily solve my problem.
>
> We have a robot with a number of different localisation systems. The aim is
> that based on estimate covariances etc, the "best" one will be used with
> some smoothing in the transition between references.
>
> Unlike the standard "map->odom->base_link->stuff" set-up that is described
> in the navigation tutorials, where you have a node publishing the
> corrections to the odometry to give you the map, our set-up is more like a
> multi-rooted tree:
>
> map_GPS->base_link
> map_INS->base_link
> map_SLAM->base_link
> odometry->base_link
> etc, and finally, some transform that is the result of some voting or
> filtering process:
> map_fused->base_link
>
> We want to be able to easily switch between how the world looks in each of
> these frames by simply selecting any of the map_* frames as the fixed frame
> in rviz and in other nodes.
>
> It seems that tf doesn't let you have a set-up like this with multiple
> roots. Any suggestions on how we should re-cast our problem to fit into the
> tf framework?
>
>
The directionality of a tf edge doesn't actually matter for most things.  So
one solution would be to just have all those map frames be children of
base_link, or of odometry (which would allow the transforms to be published
less often assuming odometry drifts fairly slowly).
- Bhaskara










> --
> View this message in context:
> http://ros-users.122217.n3.nabble.com/tf-with-multiple-localisers-tp1778461p1778461.html
> Sent from the ROS-Users mailing list archive at Nabble.com.
> _______________________________________________
> ros-users mailing list
> ros-users at code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users
>



-- 
Bhaskara Marthi
Research Scientist
Willow Garage Inc.
650-475-2856
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ros.org/pipermail/ros-users/attachments/20101027/c76f3043/attachment-0003.html>


More information about the ros-users mailing list