[ros-users] collision_map problems

Fri May 21 15:21:25 UTC 2010

Hi Ben,

A bit of cynicism never hurt anybody.  In terms of maturity all of the
arm_navigation packages are not at the point of our 1.x released packages
like 2d navigation, which have gone through a more extensive use case review
process and now have 20+ external users.  Y'all are definitely in the first
wave of folks trying to use the arm_navigation stacks and thus will
doubtless experience the perils generally associated with alpha testing.
It's very helpful to us to be reminded of things like derelict and confusing
launch files left in packages or anything else that makes the experience of
getting up to speed more difficult.  So don't hesitate to ask questions to
ros-users, to email us directly, or to submit trak tickets.

To follow up on a general point that Adolfo made, I would be wary of trying
to use any of our launch files in your own system.  The distinction we've
tried to make is that anything pr2-specific should live entirely inside of
stacks labeled with the pr2 prefix.  If you are finding things that seem pr2
specific outside of pr2 prefixed stacks we'd definitely like to know about
it.  And anything that's labeled pr2 - launch files, config files, or our
specific pr2 implementations of kinematics - is probably not going to be
that directly useful if you are trying to get things working on other
robots.  It's definitely the case at this point that it takes a good deal of
work to get all of this going on a new robot - the number of complicated
launch files in pr2_arm_navigation can attest to that.  And at this point
the tutorials are really geared more towards understanding what the packages
do in the context of a pr2 rather than being a step-by-step guide to getting
up and running with new robots.  We hope that as people are successful in
using arm_navigation that we can use their experience to write more
comprehensive tutorials and make the user experience better.

-Gil

E. Gil Jones (gjones at willowgarage.com)
Research Engineer
Willow Garage, Inc.

On Fri, May 21, 2010 at 4:05 AM, Ben Rudder <b.rudder382 at gmail.com> wrote:

> Hi all
>
> Just wanted to clarify that I wasn't being critical :) I appreciate all the
> work you guys(and girls?) are doing. It was brought to my attention that my
> previous email could be read as overly cynical, please don't take it that
> way. You've got a great system going, and really if I'm trying to use your
> stuff and have troubles, it's my own issue, haha. So yeah, please don't
> think I'm having a go at you at all! I thank you (all) a lot for your work
> and your promt replies and constant help. I hope to be able to contribute
> something useful in return at some point in time, too :)
> Cheers
>
> Ben
>
>
> On Fri, May 21, 2010 at 10:49 AM, Ben Rudder <b.rudder382 at gmail.com>wrote:
>
>> Thanks Radu.
>>
>> Yes, motion planning, towards the goal of grasping etc. Been at it for a
>> while, trying to get the stack to work with our robot, pr2 is embedded into
>> everything haha. So i thought I'd take a break and try to get something
>> related but seemingly more generic up and running, and of course I chose a
>> package that was essentially dead! Lucky me :)
>> I'll let you know how it goes.
>>
>> Ben
>>
>>
>> On Thu, May 20, 2010 at 7:43 PM, Radu Bogdan Rusu <rusu at willowgarage.com>wrote:
>>
>>> Ben,
>>>
>>> As Gil said, we're in the process of converting the code and the
>>> underlying structures to "something better" (tm). :)
>>> We're working on the details right now, but it looks like this might
>>> happen within the next few weeks, so we'll keep you
>>> posted. In the meantime, try using the current code and let us know if
>>> you have any problems with it. Is your
>>> application motion planning or something else?
>>>
>>> Cheers,
>>> Radu.
>>>
>>> On 05/20/2010 08:01 AM, Gil Jones wrote:
>>> > Hi Ben,
>>> >
>>> > You've just hit a package that needs some cleaning out.  The only
>>> > implementation of the collision map that we are currently maintaining
>>> is
>>> > collision_map_self_occ, which can do self-filtering and also maintain
>>> > collision points that are currently occluded by part of the robot.
>>> > Unfortunately the wiki documentation on this node is not really
>>> > up-to-date and the launch files and tests in this directory aren't
>>> being
>>> > maintained - we launch everything out of pr2_arm_navigation and have
>>> the
>>> > tests in arm_navigation_tests.  We also have major refactoring plans
>>> for
>>> > the node in the near future.  In the meantime, however,
>>> > pr2_arm_navigation has up-to-date launch files for this node - that's
>>> > the reason you may not have encountered any problems in the tutorials.
>>> > If you look in pr2_arm_navigation/pr2_arm_navigation_perception/launch
>>> > you'll see collision_map_self_occ.launch.  This loads a couple of yaml
>>> > files for the links to use in self-filtering and the sources of the
>>> > collision map. Let me know if you have any questions - we will be
>>> > cleaning and documenting this soon.
>>> >
>>> > -Gil
>>> >
>>> > --
>>> > E. Gil Jones (gjones at willowgarage.com <mailto:gjones at willowgarage.com
>>> >)
>>> > Research Engineer
>>> > Willow Garage, Inc.
>>> > 68 Willow Road
>>> > Menlo Park, CA 94025
>>> > 650.475.9772
>>> >
>>> >
>>> > On Thu, May 20, 2010 at 7:48 AM, Ben Rudder <b.rudder382 at gmail.com
>>> > <mailto:b.rudder382 at gmail.com>> wrote:
>>> >
>>> >     Hi all
>>> >
>>> >     I am trying to run boxturtle's collision_map_test.launch
>>> >     (collision_environment/collision_map/) and getting type errors and
>>> >     hanging,
>>> >
>>> >       [ INFO] 5.397000000: Maintaining occlusion map in frame
>>> >     'base_link', with origin at (1.100000, 0.000000, 0.000000) and
>>> >     dimension (1.000000, 1.500000, 2.000000), resolution of 0.020000;
>>> >     sensor is in frame '', fixed fame is 'base_link'.
>>> >     [ WARN] 5.407000000: Self see links need to be an array
>>> >     [ WARN] 5.425000000: No robot links will be checked for self mask
>>> >     [ WARN] 5.444000000: No links specified for self filtering.
>>> >
>>> >     (nothing happens afterwards...)
>>> >
>>> >     There's something simple going wrong, but I can't figure out how to
>>> >     pass an array as a parameter in xml for launch file to fix it?
>>> >     Newbie xml user I guess, but I also can't find it through
>>> searching.
>>> >     This test file isn't correct and the other launch file
>>> >     (collision_map.launch) has a broken link?
>>> >
>>> >     ERROR: cannot launch node of type
>>> >     [collision_map/collision_map_buffer_node]: Cannot locate node of
>>> >     type [collision_map_buffer_node] in package [collision_map]
>>> >
>>> >     despite it exisiting in the same folder.
>>> >     I'm not sure how I haven't seen this while using the move_arm
>>> >     tutorials...
>>> >
>>> >     Any Ideas/similar experiences?
>>> >
>>> >     Ben
>>> >
>>> >
>>> >     _______________________________________________
>>> >     ros-users mailing list
>>> >     ros-users at code.ros.org <mailto:ros-users at code.ros.org>
>>> >     https://code.ros.org/mailman/listinfo/ros-users
>>> >
>>> >
>>> >
>>> >
>>> > _______________________________________________
>>> > ros-users mailing list
>>> > ros-users at code.ros.org
>>> > https://code.ros.org/mailman/listinfo/ros-users
>>>
>>> --
>>> | Radu Bogdan Rusu | http://rbrusu.com/
>>> _______________________________________________
>>> ros-users mailing list
>>> ros-users at code.ros.org
>>> https://code.ros.org/mailman/listinfo/ros-users
>>>
>>
>>
>
> _______________________________________________
> ros-users mailing list
> ros-users at code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ros.org/pipermail/ros-users/attachments/20100521/b097e9d9/attachment-0003.html>