Generating IK fails for Atlas arm with ikfast 68

classic Classic list List threaded Threaded
6 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Generating IK fails for Atlas arm with ikfast 68

Stefan Kohlbrecher-2
Hi Rosen,

I just tried generating the Atlas arm IK as you stated on the mailing list this should work now with ikfast 67. I tried the latest OpenRAVE git, so version 68. Generating IK results in a 277,000 line cpp file that successfully gets generated, but compilation using gcc then fails in the last step. See attached output when trying to generate Transform6D IK.
This is on a Ubuntu 12.04 LTS/64bit machine.
Any hints for me or improvements to the generator would be appreciated :)

regards,
Stefan

------------------------------------------------------------------------------
Get your SQL database under version control now!
Version control is standard for application code, but databases havent
caught up. So what steps can you take to put your SQL databases under
version control? Why should you start doing it? Read more to find out.
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Openrave-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/openrave-users

atlas_left_arm_ikfast_68_output.txt (897K) Download Attachment
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Generating IK fails for Atlas arm with ikfast 68

Rosen Diankov
Administrator
"cc: internal compiler error:" sounds pretty serious, i'll take a look but i have a feeling that the c++ file is too big for gcc to handle.


2013/8/6 Stefan Kohlbrecher <[hidden email]>
Hi Rosen,

I just tried generating the Atlas arm IK as you stated on the mailing list this should work now with ikfast 67. I tried the latest OpenRAVE git, so version 68. Generating IK results in a 277,000 line cpp file that successfully gets generated, but compilation using gcc then fails in the last step. See attached output when trying to generate Transform6D IK.
This is on a Ubuntu 12.04 LTS/64bit machine.
Any hints for me or improvements to the generator would be appreciated :)

regards,
Stefan

------------------------------------------------------------------------------
Get your SQL database under version control now!
Version control is standard for application code, but databases havent
caught up. So what steps can you take to put your SQL databases under
version control? Why should you start doing it? Read more to find out.
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Openrave-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/openrave-users



------------------------------------------------------------------------------
Get your SQL database under version control now!
Version control is standard for application code, but databases havent
caught up. So what steps can you take to put your SQL databases under
version control? Why should you start doing it? Read more to find out.
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Openrave-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/openrave-users
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Generating IK fails for Atlas arm with ikfast 68

Stefan Kohlbrecher-2
Ok, I tried again just appending the --usecached option (to prevent re-generating the cpp file and skipping to compilation directly) and after an hour or so it compiled without error this time. My test machine was a Core2Duo with only has 4GB of RAM, so this might have contributed to the error. For further testing, I'm now trying a i7 with 16GB RAM.

Nearly 300.000 LOC for the generated cpp also seemed quite a lot to me. Is this an artifact of how this arm is structured and could be reduced, or is it normal?


2013/8/6 Rosen Diankov <[hidden email]>
"cc: internal compiler error:" sounds pretty serious, i'll take a look but i have a feeling that the c++ file is too big for gcc to handle.


2013/8/6 Stefan Kohlbrecher <[hidden email]>
Hi Rosen,

I just tried generating the Atlas arm IK as you stated on the mailing list this should work now with ikfast 67. I tried the latest OpenRAVE git, so version 68. Generating IK results in a 277,000 line cpp file that successfully gets generated, but compilation using gcc then fails in the last step. See attached output when trying to generate Transform6D IK.
This is on a Ubuntu 12.04 LTS/64bit machine.
Any hints for me or improvements to the generator would be appreciated :)

regards,
Stefan

------------------------------------------------------------------------------
Get your SQL database under version control now!
Version control is standard for application code, but databases havent
caught up. So what steps can you take to put your SQL databases under
version control? Why should you start doing it? Read more to find out.
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Openrave-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/openrave-users




------------------------------------------------------------------------------
Get your SQL database under version control now!
Version control is standard for application code, but databases havent
caught up. So what steps can you take to put your SQL databases under
version control? Why should you start doing it? Read more to find out.
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Openrave-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/openrave-users
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Generating IK fails for Atlas arm with ikfast 68

Rosen Diankov
Administrator
i'm not sure, we a lot of new degenerate conditions are now being handled. how is the performance of the IK? runtime?


2013/8/6 Stefan Kohlbrecher <[hidden email]>
Ok, I tried again just appending the --usecached option (to prevent re-generating the cpp file and skipping to compilation directly) and after an hour or so it compiled without error this time. My test machine was a Core2Duo with only has 4GB of RAM, so this might have contributed to the error. For further testing, I'm now trying a i7 with 16GB RAM.

Nearly 300.000 LOC for the generated cpp also seemed quite a lot to me. Is this an artifact of how this arm is structured and could be reduced, or is it normal?


2013/8/6 Rosen Diankov <[hidden email]>
"cc: internal compiler error:" sounds pretty serious, i'll take a look but i have a feeling that the c++ file is too big for gcc to handle.


2013/8/6 Stefan Kohlbrecher <[hidden email]>
Hi Rosen,

I just tried generating the Atlas arm IK as you stated on the mailing list this should work now with ikfast 67. I tried the latest OpenRAVE git, so version 68. Generating IK results in a 277,000 line cpp file that successfully gets generated, but compilation using gcc then fails in the last step. See attached output when trying to generate Transform6D IK.
This is on a Ubuntu 12.04 LTS/64bit machine.
Any hints for me or improvements to the generator would be appreciated :)

regards,
Stefan

------------------------------------------------------------------------------
Get your SQL database under version control now!
Version control is standard for application code, but databases havent
caught up. So what steps can you take to put your SQL databases under
version control? Why should you start doing it? Read more to find out.
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Openrave-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/openrave-users





------------------------------------------------------------------------------
Get your SQL database under version control now!
Version control is standard for application code, but databases havent
caught up. So what steps can you take to put your SQL databases under
version control? Why should you start doing it? Read more to find out.
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Openrave-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/openrave-users
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Generating IK fails for Atlas arm with ikfast 68

Stefan Kohlbrecher-2
Actually using the IK within OpenRAVE seems a bit problematic, as the --usecached option appears to only skip the cpp generation step. Afterwards, for me it always compiles the library, which takes half an hour even on the i7 machine I now use. Thus, every time, I execute "openrave.py [Some option to do with ikfast]", I have to wait for half an hour till I get back some result.

That being said, I performed a success rate test already (10000) and it looks pretty good:

openravepy.databases.inversekinematics: testik, success rate: 0.987200, wrong solutions: 0.000000, no solutions: 0.012800, missing solution: 0.102900


2013/8/6 Rosen Diankov <[hidden email]>
i'm not sure, we a lot of new degenerate conditions are now being handled. how is the performance of the IK? runtime?


2013/8/6 Stefan Kohlbrecher <[hidden email]>
Ok, I tried again just appending the --usecached option (to prevent re-generating the cpp file and skipping to compilation directly) and after an hour or so it compiled without error this time. My test machine was a Core2Duo with only has 4GB of RAM, so this might have contributed to the error. For further testing, I'm now trying a i7 with 16GB RAM.

Nearly 300.000 LOC for the generated cpp also seemed quite a lot to me. Is this an artifact of how this arm is structured and could be reduced, or is it normal?


2013/8/6 Rosen Diankov <[hidden email]>
"cc: internal compiler error:" sounds pretty serious, i'll take a look but i have a feeling that the c++ file is too big for gcc to handle.


2013/8/6 Stefan Kohlbrecher <[hidden email]>
Hi Rosen,

I just tried generating the Atlas arm IK as you stated on the mailing list this should work now with ikfast 67. I tried the latest OpenRAVE git, so version 68. Generating IK results in a 277,000 line cpp file that successfully gets generated, but compilation using gcc then fails in the last step. See attached output when trying to generate Transform6D IK.
This is on a Ubuntu 12.04 LTS/64bit machine.
Any hints for me or improvements to the generator would be appreciated :)

regards,
Stefan

------------------------------------------------------------------------------
Get your SQL database under version control now!
Version control is standard for application code, but databases havent
caught up. So what steps can you take to put your SQL databases under
version control? Why should you start doing it? Read more to find out.
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Openrave-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/openrave-users






------------------------------------------------------------------------------
Get your SQL database under version control now!
Version control is standard for application code, but databases havent
caught up. So what steps can you take to put your SQL databases under
version control? Why should you start doing it? Read more to find out.
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Openrave-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/openrave-users
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Generating IK fails for Atlas arm with ikfast 68

Rosen Diankov
Administrator
cool. i remember for the atlas arm, i went inside the actual C++ and modified some of the epsilons on the solvedialyticpoly8qep function, which boosted it to 99.9%+

and you can avoid the c++ generation step easy. just load the ik in python:

ikmodel=databases.inversekinematics.....
ikmodel.load()
ikmodel.testik('1000')

rosen,



2013/8/6 Stefan Kohlbrecher <[hidden email]>
Actually using the IK within OpenRAVE seems a bit problematic, as the --usecached option appears to only skip the cpp generation step. Afterwards, for me it always compiles the library, which takes half an hour even on the i7 machine I now use. Thus, every time, I execute "openrave.py [Some option to do with ikfast]", I have to wait for half an hour till I get back some result.

That being said, I performed a success rate test already (10000) and it looks pretty good:

openravepy.databases.inversekinematics: testik, success rate: 0.987200, wrong solutions: 0.000000, no solutions: 0.012800, missing solution: 0.102900


2013/8/6 Rosen Diankov <[hidden email]>
i'm not sure, we a lot of new degenerate conditions are now being handled. how is the performance of the IK? runtime?


2013/8/6 Stefan Kohlbrecher <[hidden email]>
Ok, I tried again just appending the --usecached option (to prevent re-generating the cpp file and skipping to compilation directly) and after an hour or so it compiled without error this time. My test machine was a Core2Duo with only has 4GB of RAM, so this might have contributed to the error. For further testing, I'm now trying a i7 with 16GB RAM.

Nearly 300.000 LOC for the generated cpp also seemed quite a lot to me. Is this an artifact of how this arm is structured and could be reduced, or is it normal?


2013/8/6 Rosen Diankov <[hidden email]>
"cc: internal compiler error:" sounds pretty serious, i'll take a look but i have a feeling that the c++ file is too big for gcc to handle.


2013/8/6 Stefan Kohlbrecher <[hidden email]>
Hi Rosen,

I just tried generating the Atlas arm IK as you stated on the mailing list this should work now with ikfast 67. I tried the latest OpenRAVE git, so version 68. Generating IK results in a 277,000 line cpp file that successfully gets generated, but compilation using gcc then fails in the last step. See attached output when trying to generate Transform6D IK.
This is on a Ubuntu 12.04 LTS/64bit machine.
Any hints for me or improvements to the generator would be appreciated :)

regards,
Stefan

------------------------------------------------------------------------------
Get your SQL database under version control now!
Version control is standard for application code, but databases havent
caught up. So what steps can you take to put your SQL databases under
version control? Why should you start doing it? Read more to find out.
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Openrave-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/openrave-users







------------------------------------------------------------------------------
Get your SQL database under version control now!
Version control is standard for application code, but databases havent
caught up. So what steps can you take to put your SQL databases under
version control? Why should you start doing it? Read more to find out.
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Openrave-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/openrave-users
Loading...