I setup a little robot using a roomba and a hokuyo lrf to do the gmapping slam, I did not use amcl at all.  You can examine my launch files and config files on github to see what I did.

https://github.com/wjwwood/sparcexplorer/blob/master/stacks/sparcexplorer/sparcexplorer_launch/launch/exploration.launch

If I wanted to have gmapping run on my laptop (if say the robot's computer was too slow) I would just comment it out of the exploration launch file and run it manually on my laptop instead.

I am also using Bosch's exploration node so the robot drives itself around exploring the area. (http://www.ros.org/wiki/exploration)

When I was first experimenting with the nav stack I got the same turning in place behavior, it was because the point you selected as a goal is off the map, which is probably the case because you have the map misconfigured or something similar.  I think if you get your move_base setup working correctly you'll find this behavior might away.

Hope that helps,

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
William Woodall
Graduate Software Engineering
Auburn University
w@auburn.edu
wjwwood@gmail.com
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~


On Wed, Jan 5, 2011 at 10:37 AM, Steven Bellens <steven.bellens@mech.kuleuven.be> wrote:
Hey,

2011/1/4 Dan Lazewatsky <dlaz@cse.wustl.edu>:
> Hi Steven -
> I've noticed similar problems on a few of our robots, mainly when using
> image_transport to send video over wifi. To figure out if wifi is the
> problem, there are two things I've found helpful: before starting any of the
> code running, from your workstation, ping the robot and see if the ping time
> increases. The other thing you can try is watching iwconfig on the robot to
> see if the link quality deteriorates over time.
> Unfortunately, I never really figured out what was going on. Replacing the
> wifi cards in the affected computers seemed to solve the problem.
> -Dan

I've figured out the problem occurs the moment I start running the
navigation stack. I've been able to build a map using gmapping
(running the gmapping mode on the groundstation), and the ping
responses were constantly okay. When I start up navigation without a
pre-build map (amcl - move base - gmapping), the ping responses
suddenly increase (from 2 - 8 ms to 1000+ ms). When I try the same
with a static map, everything goes well. Am I doing something wrong in
this setup, e.g. can I run amcl and gmapping at the same time? If not,
how should I run the navigation stack without a prebuild map?

Another problem I face is about the navigation: the holonomic robot
localizes itself OK, but the goals I send with rviz only manage to let
it rotate (there is never a translational velocity command), so the
robot starts rotating, but never gets anywhere close where I want him
to go :). Any ideas on that?

regards,

Steven

regards,

Steven

>
> On Tue, Jan 4, 2011 at 9:15 AM, Steven Bellens
> <steven.bellens@mech.kuleuven.be> wrote:
>>
>> Hi,
>>
>> when using ROS to command a mobile base, I'm running a hokuyo_node and
>> an orocos component node on the mobile base. On my workstation I have
>> the navigation stack running, which gets laserscan data from the
>> hokuyo node and sends velocity commands to the orocos component. I
>> only switched to use the wireless network recently and noticed that,
>> after running the app for some minutes, it looks like some delay
>> starts to build up: when I drive the robot with the teleop_keyboard,
>> after 2 minutes, it starts reacting very slow on my input commands.
>> What's the best way to determine if the Wifi is indeed the problem
>> and, if so, can I run this more optimally with ROS considering this
>> restriction?
>>
>> regards,
>>
>> Steven
>> _______________________________________________
>> ros-users mailing list
>> ros-users@code.ros.org
>> https://code.ros.org/mailman/listinfo/ros-users
>
>
> _______________________________________________
> ros-users mailing list
> ros-users@code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users
>
>
_______________________________________________
ros-users mailing list
ros-users@code.ros.org
https://code.ros.org/mailman/listinfo/ros-users