## playerstage-gazebo

 [PlayerStage-Gazebo] Position/Odometry very wrong very quick From: Mike Gauthier - 2005-05-12 21:21:20 ```Hello everyone, I am wondering what the best way is to keep track of a robots position using his odometry data. Right now I am using the PositionProxy and scaling the rotation data to between PI and -PI. When I run a simulation I notice that the x, y and rotation values very quickly become very wrong (I compare to the TruthProxy data). I don't want to rely on the TruthProxy for pose data because I want to run my code on real pioneer and in Gazebo. Has anyone has this same problem and found a good solution? Thanks. -Mike -- Mike Gauthier E-mail: umgaut03@... Department of Computer Science URL : http://home.cc.umanitoba.ca/~umgaut03 University of Manitoba Winnipeg, Manitoba R3T 2N2 Canada ------------------------------------------------- This mail sent through IMP: http://horde.org/imp/ ```
 Re: [PlayerStage-Gazebo] Position/Odometry very wrong very quick From: Nate Koenig - 2005-05-12 22:32:44 ```Mike, Please, please, please search the mailing lists before posting a question. Your question has been answered before: http://sourceforge.net/mailarchive/message.php?msg_id=3D11586650 Odometric data is inherently "bad", in that it drifts. You'll find that this drift is even more pronounced with real robots. So consider the drift in Gazebo a very good thing, because it forces you to deal with the problem up front. -nate On 5/12/05, Mike Gauthier wrote: > Hello everyone, >=20 > I am wondering what the best way is to keep track of a robots position us= ing his > odometry data. Right now I am using the PositionProxy and scaling the ro= tation > data to between PI and -PI. When I run a simulation I notice that the x,= y and > rotation values very quickly become very wrong (I compare to the TruthPro= xy > data). I don't want to rely on the TruthProxy for pose data because I wa= nt to > run my code on real pioneer and in Gazebo. Has anyone has this same prob= lem and > found a good solution? Thanks. >=20 > -Mike >=20 > -- >=20 > Mike Gauthier E-mail: umgaut03@... > Department of Computer Science URL : http://home.cc.umanitoba.ca/= ~umgaut03 > University of Manitoba > Winnipeg, Manitoba R3T 2N2 Canada >=20 > ------------------------------------------------- > This mail sent through IMP: http://horde.org/imp/ >=20 > ------------------------------------------------------- > This SF.Net email is sponsored by Oracle Space Sweepstakes > Want to be the first software developer in space? > Enter now for the Oracle Space Sweepstakes! > http://ads.osdn.com/?ad_id=3D7393&alloc_id=3D16281&op=3Dclick > _______________________________________________ > Playerstage-gazebo mailing list > Playerstage-gazebo@... > https://lists.sourceforge.net/lists/listinfo/playerstage-gazebo > ```
 RE: [PlayerStage-Gazebo] Position/Odometry very wrong very quick From: Mike - 2005-05-13 05:10:47 ```Thanks for the quick reply. I actually saw that post already and = thought there might be another solution. The only reason that I posted the = question is that in gazebo 0.4 I used the odometry data in simulation and on a = real pioneer. With gazebo 0.4 the error in pose was worse on the real robot compared to the simulation, but I am finding that with gazebo 0.51 the = rate at which the odometric data drifts and develops error is significantly = much quicker than it was with the older version. It's even quicker than I remember it being with the real pioneer. Do you find the same thing = when comparing gazebo 0.4 to the newest version or am I just imagining = things? Thanks again for the quick reply. -Mike -----Original Message----- From: playerstage-gazebo-admin@... [mailto:playerstage-gazebo-admin@...] On Behalf Of = Nate Koenig Sent: May 12, 2005 5:26 PM To: playerstage-gazebo@... Subject: Re: [PlayerStage-Gazebo] Position/Odometry very wrong very = quick Mike, Please, please, please search the mailing lists before posting a question. Your question has been answered before: http://sourceforge.net/mailarchive/message.php?msg_id=3D11586650 Odometric data is inherently "bad", in that it drifts. You'll find that this drift is even more pronounced with real robots. So consider the drift in Gazebo a very good thing, because it forces you to deal with the problem up front. -nate On 5/12/05, Mike Gauthier wrote: > Hello everyone, >=20 > I am wondering what the best way is to keep track of a robots position using his > odometry data. Right now I am using the PositionProxy and scaling the rotation > data to between PI and -PI. When I run a simulation I notice that the = x, y and > rotation values very quickly become very wrong (I compare to the TruthProxy > data). I don't want to rely on the TruthProxy for pose data because I want to > run my code on real pioneer and in Gazebo. Has anyone has this same problem and > found a good solution? Thanks. >=20 > -Mike >=20 > -- >=20 > Mike Gauthier E-mail: umgaut03@... > Department of Computer Science URL : http://home.cc.umanitoba.ca/~umgaut03 > University of Manitoba > Winnipeg, Manitoba R3T 2N2 Canada >=20 > ------------------------------------------------- > This mail sent through IMP: http://horde.org/imp/ >=20 > ------------------------------------------------------- > This SF.Net email is sponsored by Oracle Space Sweepstakes > Want to be the first software developer in space? > Enter now for the Oracle Space Sweepstakes! > http://ads.osdn.com/?ad_id=3D7393&alloc_id=3D16281&op=3Dclick > _______________________________________________ > Playerstage-gazebo mailing list > Playerstage-gazebo@... > https://lists.sourceforge.net/lists/listinfo/playerstage-gazebo > ------------------------------------------------------- This SF.Net email is sponsored by Oracle Space Sweepstakes Want to be the first software developer in space? Enter now for the Oracle Space Sweepstakes! http://ads.osdn.com/?ad_ids93&alloc_id=16281&op=3Dick _______________________________________________ Playerstage-gazebo mailing list Playerstage-gazebo@... https://lists.sourceforge.net/lists/listinfo/playerstage-gazebo ```
 Re: [PlayerStage-Gazebo] Position/Odometry very wrong very quick From: Stefanie Tellex - 2005-05-31 18:56:09 ```If you're using player, you could try the lodo (laser stabalized odemetry) driver here: http://robotics.usc.edu/~ahoward/pmap/, although I haven't measured how much of a difference it makes. I also noticed that odemetry error was building up quickly, so quickly that I couldn't use the odemetric angle to turn around 180 degrees in place. (The odometry would report that it had turned 180 degrees when it had really only turned about 90.) I hacked around it by dividing the amount it claimed to have turned from the starting angle by 1.5... Stefanie Mike wrote: >Thanks for the quick reply. I actually saw that post already and thought >there might be another solution. The only reason that I posted the question >is that in gazebo 0.4 I used the odometry data in simulation and on a real >pioneer. With gazebo 0.4 the error in pose was worse on the real robot >compared to the simulation, but I am finding that with gazebo 0.51 the rate >at which the odometric data drifts and develops error is significantly much >quicker than it was with the older version. It's even quicker than I >remember it being with the real pioneer. Do you find the same thing when >comparing gazebo 0.4 to the newest version or am I just imagining things? >Thanks again for the quick reply. > >-Mike >-----Original Message----- >From: playerstage-gazebo-admin@... >[mailto:playerstage-gazebo-admin@...] On Behalf Of Nate >Koenig >Sent: May 12, 2005 5:26 PM >To: playerstage-gazebo@... >Subject: Re: [PlayerStage-Gazebo] Position/Odometry very wrong very quick > >Mike, > >Please, please, please search the mailing lists before posting a >question. Your question has been answered before: > >http://sourceforge.net/mailarchive/message.php?msg_id=11586650 > >Odometric data is inherently "bad", in that it drifts. You'll find >that this drift is even more pronounced with real robots. So consider >the drift in Gazebo a very good thing, because it forces you to deal >with the problem up front. > >-nate > > >On 5/12/05, Mike Gauthier wrote: > > >>Hello everyone, >> >>I am wondering what the best way is to keep track of a robots position >> >> >using his > > >>odometry data. Right now I am using the PositionProxy and scaling the >> >> >rotation > > >>data to between PI and -PI. When I run a simulation I notice that the x, >> >> >y and > > >>rotation values very quickly become very wrong (I compare to the >> >> >TruthProxy > > >>data). I don't want to rely on the TruthProxy for pose data because I >> >> >want to > > >>run my code on real pioneer and in Gazebo. Has anyone has this same >> >> >problem and > > >>found a good solution? Thanks. >> >>-Mike >> >>-- >> >>Mike Gauthier E-mail: umgaut03@... >>Department of Computer Science URL : >> >> >http://home.cc.umanitoba.ca/~umgaut03 > > >>University of Manitoba >>Winnipeg, Manitoba R3T 2N2 Canada >> >>------------------------------------------------- >>This mail sent through IMP: http://horde.org/imp/ >> >>------------------------------------------------------- >>This SF.Net email is sponsored by Oracle Space Sweepstakes >>Want to be the first software developer in space? >>Enter now for the Oracle Space Sweepstakes! >>http://ads.osdn.com/?ad_id=7393&alloc_id=16281&op=click >>_______________________________________________ >>Playerstage-gazebo mailing list >>Playerstage-gazebo@... >>https://lists.sourceforge.net/lists/listinfo/playerstage-gazebo >> >> >> > > >------------------------------------------------------- >This SF.Net email is sponsored by Oracle Space Sweepstakes >Want to be the first software developer in space? >Enter now for the Oracle Space Sweepstakes! >http://ads.osdn.com/?ad_ids93&alloc_id281&op=ick >_______________________________________________ >Playerstage-gazebo mailing list >Playerstage-gazebo@... >https://lists.sourceforge.net/lists/listinfo/playerstage-gazebo > > > >------------------------------------------------------- >This SF.Net email is sponsored by Oracle Space Sweepstakes >Want to be the first software developer in space? >Enter now for the Oracle Space Sweepstakes! >http://ads.osdn.com/?ad_ids93&alloc_id281&op=click >_______________________________________________ >Playerstage-gazebo mailing list >Playerstage-gazebo@... >https://lists.sourceforge.net/lists/listinfo/playerstage-gazebo > > ```
 Re: [PlayerStage-Gazebo] Position/Odometry very wrong very quick From: Mike Gauthier - 2005-05-31 22:10:29 ```Ya, my problem ended up being conveniently solved because I was using the PioneerAT model in my world rather than what I wanted to be using, the PioneerDX. This probably still means that there is something up with the PioneerAT model but i'm not too worried about looking into it any more than I already did. I was also dividing the rotation by something like 1.5 which was pretty much dead on but the x and y positions were still so unpredictable, although my hack wasn't directly to the gazebo source, just in my client. Quoting Stefanie Tellex : > If you're using player, you could try the lodo (laser stabalized > odemetry) driver here: http://robotics.usc.edu/~ahoward/pmap/, although > I haven't measured how much of a difference it makes. > > I also noticed that odemetry error was building up quickly, so quickly > that I couldn't use the odemetric angle to turn around 180 degrees in > place. (The odometry would report that it had turned 180 degrees when it > had really only turned about 90.) I hacked around it by dividing the > amount it claimed to have turned from the starting angle by 1.5... > > Stefanie > > Mike wrote: > > >Thanks for the quick reply. I actually saw that post already and thought > >there might be another solution. The only reason that I posted the > question > >is that in gazebo 0.4 I used the odometry data in simulation and on a real > >pioneer. With gazebo 0.4 the error in pose was worse on the real robot > >compared to the simulation, but I am finding that with gazebo 0.51 the > rate > >at which the odometric data drifts and develops error is significantly > much > >quicker than it was with the older version. It's even quicker than I > >remember it being with the real pioneer. Do you find the same thing when > >comparing gazebo 0.4 to the newest version or am I just imagining things? > >Thanks again for the quick reply. > > > >-Mike > >-----Original Message----- > >From: playerstage-gazebo-admin@... > >[mailto:playerstage-gazebo-admin@...] On Behalf Of Nate > >Koenig > >Sent: May 12, 2005 5:26 PM > >To: playerstage-gazebo@... > >Subject: Re: [PlayerStage-Gazebo] Position/Odometry very wrong very quick > > > >Mike, > > > >Please, please, please search the mailing lists before posting a > >question. Your question has been answered before: > > > >http://sourceforge.net/mailarchive/message.php?msg_id=11586650 > > > >Odometric data is inherently "bad", in that it drifts. You'll find > >that this drift is even more pronounced with real robots. So consider > >the drift in Gazebo a very good thing, because it forces you to deal > >with the problem up front. > > > >-nate > > > > > >On 5/12/05, Mike Gauthier wrote: > > > > > >>Hello everyone, > >> > >>I am wondering what the best way is to keep track of a robots position > >> > >> > >using his > > > > > >>odometry data. Right now I am using the PositionProxy and scaling the > >> > >> > >rotation > > > > > >>data to between PI and -PI. When I run a simulation I notice that the x, > >> > >> > >y and > > > > > >>rotation values very quickly become very wrong (I compare to the > >> > >> > >TruthProxy > > > > > >>data). I don't want to rely on the TruthProxy for pose data because I > >> > >> > >want to > > > > > >>run my code on real pioneer and in Gazebo. Has anyone has this same > >> > >> > >problem and > > > > > >>found a good solution? Thanks. > >> > >>-Mike > >> > >>-- > >> > >>Mike Gauthier E-mail: umgaut03@... > >>Department of Computer Science URL : > >> > >> > >http://home.cc.umanitoba.ca/~umgaut03 > > > > > >>University of Manitoba > >>Winnipeg, Manitoba R3T 2N2 Canada > >> > >>------------------------------------------------- > >>This mail sent through IMP: http://horde.org/imp/ > >> > >>------------------------------------------------------- > >>This SF.Net email is sponsored by Oracle Space Sweepstakes > >>Want to be the first software developer in space? > >>Enter now for the Oracle Space Sweepstakes! > >>http://ads.osdn.com/?ad_id=7393&alloc_id=16281&op=click > >>_______________________________________________ > >>Playerstage-gazebo mailing list > >>Playerstage-gazebo@... > >>https://lists.sourceforge.net/lists/listinfo/playerstage-gazebo > >> > >> > >> > > > > > >------------------------------------------------------- > >This SF.Net email is sponsored by Oracle Space Sweepstakes > >Want to be the first software developer in space? > >Enter now for the Oracle Space Sweepstakes! > >http://ads.osdn.com/?ad_ids93&alloc_id281&op=ick > >_______________________________________________ > >Playerstage-gazebo mailing list > >Playerstage-gazebo@... > >https://lists.sourceforge.net/lists/listinfo/playerstage-gazebo > > > > > > > >------------------------------------------------------- > >This SF.Net email is sponsored by Oracle Space Sweepstakes > >Want to be the first software developer in space? > >Enter now for the Oracle Space Sweepstakes! > >http://ads.osdn.com/?ad_ids93&alloc_id281&op=click > >_______________________________________________ > >Playerstage-gazebo mailing list > >Playerstage-gazebo@... > >https://lists.sourceforge.net/lists/listinfo/playerstage-gazebo > > > > > > > > ------------------------------------------------------- > This SF.Net email is sponsored by Yahoo. > Introducing Yahoo! Search Developer Network - Create apps using Yahoo! > Search APIs Find out how you can build Yahoo! directly into your own > Applications - visit http://developer.yahoo.net/?fr=offad-ysdn-ostg-q22005 > _______________________________________________ > Playerstage-gazebo mailing list > Playerstage-gazebo@... > https://lists.sourceforge.net/lists/listinfo/playerstage-gazebo > -- Mike Gauthier E-mail: umgaut03@... Department of Computer Science URL : http://home.cc.umanitoba.ca/~umgaut03 University of Manitoba Winnipeg, Manitoba R3T 2N2 Canada ------------------------------------------------- This mail sent through IMP: http://horde.org/imp/ ```
 Re: [PlayerStage-Gazebo] Position/Odometry very wrong very quick From: David Cameron - 2005-06-01 01:54:30 ```In the PioneerAT model, the x and y positions are estimated from the angle and the travel distance. ( x =3D r sin(theta), y =3D r cos(theta) type stuff.) If you're correcting the angle in your client, then it won't be right in the x and y calculations in the simulator. Changing the models isn't too bad, I've done a fair bit to the CarChassis model, so if either of you would like to improve the telemetry, I'd recommend checking them out. If you want to transfer your code to a real robot, it might be a more natural way of doing things. Dave On 5/31/05, Mike Gauthier wrote: > Ya, my problem ended up being conveniently solved because I was using the > PioneerAT model in my world rather than what I wanted to be using, the > PioneerDX. This probably still means that there is something up with the > PioneerAT model but i'm not too worried about looking into it any more th= an I > already did. I was also dividing the rotation by something like 1.5 whic= h was > pretty much dead on but the x and y positions were still so unpredictable= , > although my hack wasn't directly to the gazebo source, just in my client. >=20 > Quoting Stefanie Tellex : >=20 > > If you're using player, you could try the lodo (laser stabalized > > odemetry) driver here: http://robotics.usc.edu/~ahoward/pmap/, although > > I haven't measured how much of a difference it makes. > > > > I also noticed that odemetry error was building up quickly, so quickly > > that I couldn't use the odemetric angle to turn around 180 degrees in > > place. (The odometry would report that it had turned 180 degrees when i= t > > had really only turned about 90.) I hacked around it by dividing the > > amount it claimed to have turned from the starting angle by 1.5... > > > > Stefanie > > > > Mike wrote: > > > > >Thanks for the quick reply. I actually saw that post already and thou= ght > > >there might be another solution. The only reason that I posted the > > question > > >is that in gazebo 0.4 I used the odometry data in simulation and on a = real > > >pioneer. With gazebo 0.4 the error in pose was worse on the real robo= t > > >compared to the simulation, but I am finding that with gazebo 0.51 the > > rate > > >at which the odometric data drifts and develops error is significantly > > much > > >quicker than it was with the older version. It's even quicker than I > > >remember it being with the real pioneer. Do you find the same thing w= hen > > >comparing gazebo 0.4 to the newest version or am I just imagining thin= gs? > > >Thanks again for the quick reply. > > > > > >-Mike > > >-----Original Message----- > > >From: playerstage-gazebo-admin@... > > >[mailto:playerstage-gazebo-admin@...] On Behalf Of N= ate > > >Koenig > > >Sent: May 12, 2005 5:26 PM > > >To: playerstage-gazebo@... > > >Subject: Re: [PlayerStage-Gazebo] Position/Odometry very wrong very qu= ick > > > > > >Mike, > > > > > >Please, please, please search the mailing lists before posting a > > >question. Your question has been answered before: > > > > > >http://sourceforge.net/mailarchive/message.php?msg_id=3D11586650 > > > > > >Odometric data is inherently "bad", in that it drifts. You'll find > > >that this drift is even more pronounced with real robots. So consider > > >the drift in Gazebo a very good thing, because it forces you to deal > > >with the problem up front. > > > > > >-nate > > > > > > > > >On 5/12/05, Mike Gauthier wrote: > > > > > > > > >>Hello everyone, > > >> > > >>I am wondering what the best way is to keep track of a robots positio= n > > >> > > >> > > >using his > > > > > > > > >>odometry data. Right now I am using the PositionProxy and scaling th= e > > >> > > >> > > >rotation > > > > > > > > >>data to between PI and -PI. When I run a simulation I notice that th= e x, > > >> > > >> > > >y and > > > > > > > > >>rotation values very quickly become very wrong (I compare to the > > >> > > >> > > >TruthProxy > > > > > > > > >>data). I don't want to rely on the TruthProxy for pose data because = I > > >> > > >> > > >want to > > > > > > > > >>run my code on real pioneer and in Gazebo. Has anyone has this same > > >> > > >> > > >problem and > > > > > > > > >>found a good solution? Thanks. > > >> > > >>-Mike > > >> > > >>-- > > >> > > >>Mike Gauthier E-mail: umgaut03@... > > >>Department of Computer Science URL : > > >> > > >> > > >http://home.cc.umanitoba.ca/~umgaut03 > > > > > > > > >>University of Manitoba > > >>Winnipeg, Manitoba R3T 2N2 Canada > > >> > > >>------------------------------------------------- > > >>This mail sent through IMP: http://horde.org/imp/ > > >> > > >>------------------------------------------------------- > > >>This SF.Net email is sponsored by Oracle Space Sweepstakes > > >>Want to be the first software developer in space? > > >>Enter now for the Oracle Space Sweepstakes! > > >>http://ads.osdn.com/?ad_id=3D7393&alloc_id=3D16281&op=3Dclick > > >>_______________________________________________ > > >>Playerstage-gazebo mailing list > > >>Playerstage-gazebo@... > > >>https://lists.sourceforge.net/lists/listinfo/playerstage-gazebo > > >> > > >> > > >> > > > > > > > > >------------------------------------------------------- > > >This SF.Net email is sponsored by Oracle Space Sweepstakes > > >Want to be the first software developer in space? > > >Enter now for the Oracle Space Sweepstakes! > > >http://ads.osdn.com/?ad_ids93&alloc_id=16281&op=3Dick > > >_______________________________________________ > > >Playerstage-gazebo mailing list > > >Playerstage-gazebo@... > > >https://lists.sourceforge.net/lists/listinfo/playerstage-gazebo > > > > > > > > > > > >------------------------------------------------------- > > >This SF.Net email is sponsored by Oracle Space Sweepstakes > > >Want to be the first software developer in space? > > >Enter now for the Oracle Space Sweepstakes! > > >http://ads.osdn.com/?ad_ids93&alloc_id=16281&op=3Dclick > > >_______________________________________________ > > >Playerstage-gazebo mailing list > > >Playerstage-gazebo@... > > >https://lists.sourceforge.net/lists/listinfo/playerstage-gazebo > > > > > > > > > > > > > > ------------------------------------------------------- > > This SF.Net email is sponsored by Yahoo. > > Introducing Yahoo! Search Developer Network - Create apps using Yahoo! > > Search APIs Find out how you can build Yahoo! directly into your own > > Applications - visit http://developer.yahoo.net/?fr=3Doffad-ysdn-ostg-q= 22005 > > _______________________________________________ > > Playerstage-gazebo mailing list > > Playerstage-gazebo@... > > https://lists.sourceforge.net/lists/listinfo/playerstage-gazebo > > >=20 >=20 > -- >=20 > Mike Gauthier E-mail: umgaut03@... > Department of Computer Science URL : http://home.cc.umanitoba.ca/= ~umgaut03 > University of Manitoba > Winnipeg, Manitoba R3T 2N2 Canada >=20 > ------------------------------------------------- > This mail sent through IMP: http://horde.org/imp/ >=20 >=20 > ------------------------------------------------------- > This SF.Net email is sponsored by Yahoo. > Introducing Yahoo! Search Developer Network - Create apps using Yahoo! > Search APIs Find out how you can build Yahoo! directly into your own > Applications - visit http://developer.yahoo.net/?fr=3Doffad-ysdn-ostg-q22= 005 > _______________________________________________ > Playerstage-gazebo mailing list > Playerstage-gazebo@... > https://lists.sourceforge.net/lists/listinfo/playerstage-gazebo >=20 --=20 David Cameron University of British Columbia http://bar.psych.ubc.ca/People/Dave.html ```