[ros-users] ND, VHF etc in ROS ?

[Gonçalo Cabrita]

Hi Brian and Arkapravo,

Here at the lab we also went from Player to ROS. In the beginning it looked like that the navigation stack was too heavy to run on our robots, so we began porting ND to ROS. We quickly fell in love with the navigation stack and dropped that idea. The navigation stack is highly customizable and can fit the needs of a wide range of robots.

I would start by checking the navigation stack as Brian suggested and move on from there.

Regards,

Gonçalo Cabrita
ISR University of Coimbra
Portugal

On May 1, 2011, at 9:06 PM, Brian Gerkey wrote:

> On Fri, Apr 29, 2011 at 11:53 PM, Arkapravo Bhaumik
> <arkapravobhaumik at gmail.com> wrote:
>> My question, Player Project (Player Stage in particular) which supported
>> popular navigation algorithms (viz. VFH, ND, line/wall following, SND etc),
>> are these available in ROS ?
> 
> hi Arkapravo,
> 
> There's a flexible and nicely-featured navigation stack in ROS:
>  http://www.ros.org/wiki/navigation
> It provides the pieces necessary to make a planar robot navigate, with
> or without a map.
> 
> I haven't seen an implementation of VFH (in this context, Vector Field
> Histogram; not to be confused with Viewpoint Feature Histogram) in
> ROS.  Same for ND and SND.  But unless you're looking for those
> algorithms specifically, you should have a look at the
> base_local_planner, which implements the Trajectory Rollout and DWA
> algorithms for local goal-seeking obstacle avoidance:
>  http://www.ros.org/wiki/base_local_planner
> There's a nicer implementation of DWA in dwa_local_planner:
>  http://www.ros.org/wiki/dwa_local_planner
> 
> A number of drivers from Player have been brought into ROS.  Some
> notes on how to do it are here:
>  http://www.ros.org/wiki/ROS/PlayerIntegration
> I would expect the ND driver to be easily ported from Player to ROS,
> because underneath it's a standalone library.  Porting VFH would be
> unpleasant, because the implementation is very Player-specific.  I'm
> not familiar enough with SND to say how much work it would be.
> 
> 	brian.
> _______________________________________________
> ros-users mailing list
> ros-users at code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users