Eitan,
I copied nav_view.xml from move_base_stage/move_base, changing the
remappings to topics that exist in my system. That along with your two
suggestions seem to have done the trick. There are, however, a few odd
things:
-when I start up nav_view, the map takes a *really* long time to show up
- on the order of 5 minutes
-once the robot gets to its goal, it spins around a lot before stopping
-nav_view seems a bit unstable - ubuntu frequently grays it out thinking
it's crashed
Thanks for the help,
-Dan
On 3/31/10 12:24 PM, Eitan Marder-Eppstein wrote:
> Dan,
>
> The Costmap2DROS transform timeouts could be a problem if they don't
> taper off after startup. A bunch of warnings right as you start isn't
> unexpected, but if they just keep spewing that means there is likely a
> timing issue. Do they keep spewing constantly or more or less stop
> after a few seconds?
>
> After looking at your configuration files, it seems that you're
> missing a pretty important line in costmap_common_params.yaml. You
> define a "laser_scan_sensor," but you never actually tell the
> navigation stack to use it. As such, you'll never see obstacles show
> up in the costmap. Add the following line to your
> costmap_common_params.yaml file:
>
> observation_sources: laser_scan_sensor
>
> This will tell the navigation stack that you'd like to use the
> laser_scan_sensor you defined.
>
> Hopefully, adding that line should get obstacles to show up for you.
> However, even without that, I would've expected the robot to move when
> you set a goal, it would just hit stuff. Are you sure that you were
> sending a goal on the correct topic? In rviz, you need to make sure to
> set the 2D Nav Goal topic to "move_base_simple/goal." I believe that
> both rviz and nav_view by default send to the "goal" topic instead. If
> you want to use nav_view, you'll have to remap in your launch file.
> You can see move_base_stage/move_base/nav_view.xml for an example of this.
>
> Oh... I think I may know why the robot doesn't move, you've defined
> your footprint by only 2 points. The navigation stack supports any
> convex polygon for a footprint and, therefore, needs at least three
> points to run. It also assumes that the robot is centered at (0, 0).
> If you want to define a square robot of the dimensions in your launch
> file you'll need to use the following:
>
> footprint: [[-0.2, -0.2], [-0.2, 0.2], [0.2, 0.2], [0.2, -0.2]]
>
> You should have received a warning about this, but I just realized
> that the costmap uses ROS_ASSERT where it should be using ROS_FATAL.
> This means that all those warnings will be compiled out when run in
> release mode. I've ticketed fixing this here:
> https://code.ros.org/trac/ros-pkg/ticket/3939
>
> Hope this helps and let me know how things go... we'll figure it out yet,
>
> Eitan
>
> On Wed, Mar 31, 2010 at 8:13 AM, Dan Lazewatsky
> <lazewatskyd@cse.wustl.edu <mailto:lazewatskyd@cse.wustl.edu>> wrote:
>
>
> On 3/30/10 7:55 PM, Eitan Marder-Eppstein wrote:
>> Hey Dan,
>>
>> A few questions for ya to help figure out what's going on:
>>
>> 1) Have you tried using the rxconsole tool to see if any useful
>> information is being output from the system? If so, did you see
>> anything that seemed to be complaining or giving warnings?
> rxconsole doesn't show anything that seems too concerning. The
> worst is a bunch of
> Costmap2DROS transform timeout. Current time: ...
>
>>
>> 2) What launch files are you running on the erratic? Are they the
>> ones in the 2dnav_erratic package? Or different ones that you
>> created? We used to have an erratic in-house, but we don't
>> anymore and its been awhile since the 2dnav_erratic package has
>> been tested. If you're running with your own configuration, could
>> you post the files so I can take a look at them?
> I've attached the launch file and config files I'm using. The
> config files were all setup according to this:
> http://www.ros.org/wiki/navigation/Tutorials/RobotSetup
>>
>> 3) Do you see any obstacles in nav_view? On second thought, rviz
>> is a more powerful debugging tool for this kind of stuff. Have
>> you tried using rviz to get useful debugging information (whether
>> obstacles show up, a plan is created, etc)? You can find a
>> tutorial on using rviz with the navigation stack here:
>> http://www.ros.org/wiki/navigation/Tutorials/Using%20rviz%20with%20the%20Navigation%20Stack
> nav_view and rviz don't show any obstacles, plan, or really
> anything but the map. If I'm really lucky, rviz will show the amcl
> particle cloud.
>
>>
>>
>> Ok... take a shot at answering these and we'll see if that gives
>> a bit more insight into what's going on.
>>
>> Hope all is well,
>>
>> Eitan
>>
>> On Tue, Mar 30, 2010 at 3:57 PM, Dan Lazewatsky
>> <lazewatskyd@cse.wustl.edu <mailto:lazewatskyd@cse.wustl.edu>> wrote:
>>
>> Thanks for the response Eitan -
>> I've created a map, and am trying to get an erratic to
>> navigate in said map using nav_view. I set up the robot
>> according the the nav robot setup tutorial, and when I run
>> nav view I can see the map and set the robot's initial pose,
>> but trying to set the goal doesn't seem to do anything. I get
>> a printout from nav view saying "setting goal: Position(...)"
>> but that's all.
>>
>> -Dan
>>
>>
>> On 3/30/10 4:31 PM, Eitan Marder-Eppstein wrote:
>>> Dan,
>>>
>>> The wavefront package has fallen into a bit of disrepair as
>>> we favor our move_base package for autonomous navigation. If
>>> you're looking to run some autonomous navigation demos,
>>> check out the move_base_stage package instead of the
>>> 2dnav_stage package. The 2dnav_stage package will probably
>>> be removed in the near future to save people the confusion.
>>>
>>> Hope this helps,
>>>
>>> Eitan
>>>
>>> On Tue, Mar 30, 2010 at 12:07 PM, Dan Lazewatsky
>>> <lazewatskyd@cse.wustl.edu
>>> <mailto:lazewatskyd@cse.wustl.edu>> wrote:
>>>
>>> Hi all - I'm having some trouble using wavefront. When I
>>> try to start it I get this:
>>>
>>> robotics@tears$ rosrun wavefront wavefront
>>> terminate called after throwing an instance of 'ros::InvalidNameException'
>>> what(): Using ~ names with NodeHandle methods is not allowed. If you want to use private names with the NodeHandle interface, construct a NodeHandle using a private name as its namespace. e.g. ros::NodeHandle nh("~"); nh.getParam("my_private_name"); (name = [~dist_eps])
>>> /home/robotics/ros/ros/bin/rosrun: line 35: 15562 Aborted $exepath "$@"
>>>
>>> Granted, I don't know what the right way to use
>>> wavefront since the documentation is a bit spotty.
>>>
>>> Any ideas would be appreciated.
>>>
>>> Thanks,
>>> -Dan
>>>
>>> _______________________________________________
>>> ros-users mailing list
>>> ros-users@code.ros.org <mailto:ros-users@code.ros.org>
>>> https://code.ros.org/mailman/listinfo/ros-users
>>>
>>>
>>>
>>> _______________________________________________
>>> ros-users mailing list
>>> ros-users@code.ros.org <mailto:ros-users@code.ros.org>
>>> https://code.ros.org/mailman/listinfo/ros-users
>>>
>>
>> _______________________________________________
>> ros-users mailing list
>> ros-users@code.ros.org <mailto:ros-users@code.ros.org>
>> https://code.ros.org/mailman/listinfo/ros-users
>>
>>
>>
>> _______________________________________________
>> ros-users mailing list
>> ros-users@code.ros.org <mailto:ros-users@code.ros.org>
>> https://code.ros.org/mailman/listinfo/ros-users
>>
>
> _______________________________________________
> ros-users mailing list
> ros-users@code.ros.org <mailto: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
>
(text/plain)