From: Angel C. <aco...@gm...> - 2007-05-02 18:45:19
|
Hi everybody, is there any kind of odometry noise implemented for player (version 2.0.3) ? I am working with robot Pioneer and the simulator player and I need to use odometry noise. Thanks very much. |
From: Paul O. <new...@ki...> - 2007-05-02 20:07:02
|
On Wed, 2 May 2007, Angel Cortes wrote: > Hi everybody, > > is there any kind of odometry noise implemented for player (version 2.0.3) ? > > I am working with robot Pioneer and the simulator player and I need to use > odometry noise. > > Thanks very much. > > Good question and I'm joining to it. One of the PhD's at PJIIT keeps on askin me about it and about sonar readnigs noise too (i.e. readings in the corner should differ). Would be great to see Stage simulator more realistic! |
From: Reed H. <re...@mo...> - 2007-05-02 21:07:12
|
Paul Osmialowski wrote: >> is there any kind of odometry noise implemented for player (version 2.0.3) ? >> >> > Good question and I'm joining to it. One of the PhD's at PJIIT keeps on > askin me about it and about sonar readnigs noise too (i.e. readings in the > corner should differ). Would be great to see Stage simulator more > realistic! Stage can accumulate error in odometry, use the "odom" localization type and set odom_error property to the ranges for error in X, Y and angle (it adds a random value from within that range each update). No high frequency noise or jitter though. I made modified laser and modified ranger models that do extra stuff. My laser has a small bit of "jitter" noise in range and angle. For the ranger I added an option to test a series of positions in front of the simulated sensor, rather than just one, and pick the closest (so that's like a cone-shaped signal from a sonar transducer hitting the closest object within its field). Most sonar is actually very stable over time, but it has problems with reflection, I'd like to simulate that at some point too. (Something that's a bit like that that I also experimented a bit with is treating a sonar reading as missing if the angle of incidence with an object was too great.) Patches for these are on the patch tracker I think or I can send them to you. This kind of thing adds a lot more computation though, so I plan on switching them to be seperate model types in the future and maybe adding a thing to Stage that lets you select between different implementations of a particular model type. Reed |
From: Paul O. <new...@ki...> - 2007-05-03 08:48:57
|
> Patches for these are on the patch > tracker I think or I can send them to you. This kind of thing adds a lot more > computation though, so I plan on switching them to be seperate model types in > the future and maybe adding a thing to Stage that lets you select between > different implementations of a particular model type. Thanks for your answer Reed. If it is there, you can give me Request IDs of desired patches. Hope one day it will be a part of the mainstream Stage. Paul |