Objects sinking through floor

Discussion:

Rein Appeldoorn

2011-06-08 18:00:25 UTC

Hi,

I'm working on a simulation of an autonomous football table. When I model
the ball, it's sinking through the floor. Because the geometry of the ball
is quite small (radius of 0.015), this sinking is not acceptable. Strange
enough, when I check the other objects, they're all sinking a little bit.
Does anybody have an explanation for this?

A thing I can do is scale up all models, this makes the relative errors much
less. Does anybody got experience with this fact? Is scaling a good
solution?

Thanks in advance!

Regards,

Rein

John Hsu

2011-06-08 18:33:56 UTC

Permalink

Hi,

you can check to see if contactSurfaceLayer is small, e.g.

<contactSurfaceLayer>0</contactSurfaceLayer>

in addition, you might want to make sure <CFM> is 0.

hope this helps.

John

Post by Rein Appeldoorn
Hi,
I'm working on a simulation of an autonomous football table. When I model
the ball, it's sinking through the floor. Because the geometry of the ball
is quite small (radius of 0.015), this sinking is not acceptable. Strange
enough, when I check the other objects, they're all sinking a little bit.
Does anybody have an explanation for this?
A thing I can do is scale up all models, this makes the relative errors much
less. Does anybody got experience with this fact? Is scaling a good
solution?
Thanks in advance!
Regards,
Rein
------------------------------------------------------------------------------
EditLive Enterprise is the world's most technically advanced content
authoring tool. Experience the power of Track Changes, Inline Image
Editing and ensure content is compliant with Accessibility Checking.
http://p.sf.net/sfu/ephox-dev2dev
_______________________________________________
Playerstage-gazebo mailing list
https://lists.sourceforge.net/lists/listinfo/playerstage-gazebo

--
John Hsu, Ph.D.
Senior Research Engineer
Willow Garage Inc.
+1-650-475-2809

Rein Appeldoorn

2011-06-09 07:07:27 UTC

Permalink

I checked those two parameters but nothing changes. Is this a known fact or
is this just something weird? Here is the world file I'm using:

-----------------------------------------------------------------------------------

<?xml version="1.0"?>

<gazebo:world
xmlns:xi="http://www.w3.org/2001/XInclude"
xmlns:gazebo="http://playerstage.sourceforge.net/gazebo/xmlschema/#gz"
xmlns:model="http://playerstage.sourceforge.net/gazebo/xmlschema/#model"

xmlns:sensor="
http://playerstage.sourceforge.net/gazebo/xmlschema/#sensor"
xmlns:window="
http://playerstage.sourceforge.net/gazebo/xmlschema/#window"
xmlns:param="http://playerstage.sourceforge.net/gazebo/xmlschema/#param"

xmlns:body="http://playerstage.sourceforge.net/gazebo/xmlschema/#body"
xmlns:geom="http://playerstage.sourceforge.net/gazebo/xmlschema/#geom"
xmlns:joint="http://playerstage.sourceforge.net/gazebo/xmlschema/#joint"

xmlns:interface="
http://playerstage.sourceforge.net/gazebo/xmlschema/#interface"
xmlns:ui="http://playerstage.sourceforge.net/gazebo/xmlschema/#ui"
xmlns:rendering="
http://playerstage.sourceforge.net/gazebo/xmlschema/#rendering"
xmlns:controller="
http://playerstage.sourceforge.net/gazebo/xmlschema/#controller"
xmlns:physics="
http://playerstage.sourceforge.net/gazebo/xmlschema/#physics" >

<verbosity>5</verbosity>

<physics:ode>
<stepTime>0.005</stepTime>
<gravity>0 0 -9.8</gravity>
<cfm>0</cfm>
<erp>0.8</erp>
<quickStep>true</quickStep>
<quickStepIters>100</quickStepIters>
<quickStepW>1.3</quickStepW>
<contactMaxCorrectingVel>100.0</contactMaxCorrectingVel>
<contactSurfaceLayer>0.0</contactSurfaceLayer>
<updateRate>0</updateRate>
</physics:ode>

<rendering:gui>
<type>fltk</type>
<size>800 600</size>
<pos>0 0</pos>
</rendering:gui>

<rendering:ogre>
<ambient>0.8 0.8 0.8 1.0</ambient>
<sky>
<material>Gazebo/CloudySky</material>
</sky>
<grid>false</grid>
<shadowTechnique>stencilModulative</shadowTechnique>
</rendering:ogre>


<model:physical name="groundplane_model">
<xyz>0 0 0</xyz>
<rpy>0 0 0</rpy>
<static>true</static>

<body:plane name="ground_plane_body">
<geom:plane name="ground_plane_geom">
<normal>0 0 1</normal>
<size>1000 1000</size>
<segments>100 100</segments>
<uvTile>100 100</uvTile>
<material>Gazebo/GrassFloor</material>
<mu1>109999.0</mu1>
<mu2>1000.0</mu2>
</geom:plane>
</body:plane>
</model:physical>



<model:physical name="ball_model">
<xyz>0.025 0 1</xyz>
<static>false</static>

<body:sphere name="ball_body">
<geom:sphere name="ball_geom">
<xyz>0 0 0</xyz>
<size>0.015</size>
<mass>0.02</mass>

<visual>
<size>0.03 0.03 0.03</size>
<mesh>unit_sphere</mesh>
<material>Gazebo/White</material>
</visual>
<mu1>109999.0</mu1>
</geom:sphere>
</body:sphere>
</model:physical>



<model:renderable name="point_white">
<xyz>-3 0 5</xyz>
<static>true</static>
<light>
<type>point</type>
<diffuseColor>0.8 0.8 0.8</diffuseColor>
<specularColor>0.1 0.1 0.1</specularColor>
<range>20</range>


<attenuation>0.3 0.01 0</attenuation>
</light>
</model:renderable>


</gazebo:world>

-----------------------------------------------------------------------------------

Do you guys get the same behavior? If so, is there any fix or a way to get
rid of this?

Thanks in advance!

Rein

Post by John Hsu
Hi,
you can check to see if contactSurfaceLayer is small, e.g.
<contactSurfaceLayer>0</contactSurfaceLayer>
in addition, you might want to make sure <CFM> is 0.
hope this helps.
John

Post by Rein Appeldoorn
Hi,
I'm working on a simulation of an autonomous football table. When I model
the ball, it's sinking through the floor. Because the geometry of the

ball

Post by Rein Appeldoorn
is quite small (radius of 0.015), this sinking is not acceptable. Strange
enough, when I check the other objects, they're all sinking a little bit.
Does anybody have an explanation for this?
A thing I can do is scale up all models, this makes the relative errors much
less. Does anybody got experience with this fact? Is scaling a good
solution?
Thanks in advance!
Regards,
Rein

------------------------------------------------------------------------------

Post by Rein Appeldoorn
EditLive Enterprise is the world's most technically advanced content
authoring tool. Experience the power of Track Changes, Inline Image
Editing and ensure content is compliant with Accessibility Checking.
http://p.sf.net/sfu/ephox-dev2dev
_______________________________________________
Playerstage-gazebo mailing list
https://lists.sourceforge.net/lists/listinfo/playerstage-gazebo

--
John Hsu, Ph.D.
Senior Research Engineer
Willow Garage Inc.
+1-650-475-2809
------------------------------------------------------------------------------
EditLive Enterprise is the world's most technically advanced content
authoring tool. Experience the power of Track Changes, Inline Image
Editing and ensure content is compliant with Accessibility Checking.
http://p.sf.net/sfu/ephox-dev2dev
_______________________________________________
Playerstage-gazebo mailing list
https://lists.sourceforge.net/lists/listinfo/playerstage-gazebo

Kamil Muszynski

2011-06-09 07:38:33 UTC

Permalink

Hello,

here is a sample ball model I'm using for my simulations. Notice the
<kd> and <kp> params and try to play with their values - they control
how much force
is applied to this specific ball model to prevent it from sinking:

<body:sphere name = "ball_body">
<dampingFactor>0.0001</dampingFactor>
<xyz>0 0 0</xyz>
<rpy>0 0 0</rpy>
<geom:sphere name = "ball_geom">
<xyz>0 0 0.01</xyz>
<rpy>0 0 0 </rpy>
<size>0.02</size>
<mu1>0.3</mu1>
<kp>2e6</kp>
<kd>2e6</kd>
<bounce>120</bounce>
<bounceVel>0.5</bounceVel>

<slip1>0.7</slip1>
<slip2>0.7</slip2>

<mass>0.045</mass>
<visual>
<scale>0.02 0.02 0.02</scale>
<mesh>unit_sphere</mesh>
<material>Gazebo/Robocup_orange</material>
</visual>
</geom:sphere>
</body:sphere>

Also global parameters like step time, erp and cfm does matter, I'm
using those values:
<stepTime>0.002</stepTime>
<erp>0.9</erp>
<cfm>0.01</cfm>

I hope that will help you,

Best regards,
Kamil Muszyński

Post by Rein Appeldoorn
I checked those two parameters but nothing changes. Is this a known fact or
-----------------------------------------------------------------------------------
<?xml version="1.0"?>
<gazebo:world
xmlns:xi="http://www.w3.org/2001/XInclude"
xmlns:gazebo="http://playerstage.sourceforge.net/gazebo/xmlschema/#gz"
xmlns:model="http://playerstage.sourceforge.net/gazebo/xmlschema/#model"
xmlns:sensor="
http://playerstage.sourceforge.net/gazebo/xmlschema/#sensor"
xmlns:window="
http://playerstage.sourceforge.net/gazebo/xmlschema/#window"
xmlns:param="http://playerstage.sourceforge.net/gazebo/xmlschema/#param"
xmlns:body="http://playerstage.sourceforge.net/gazebo/xmlschema/#body"
xmlns:geom="http://playerstage.sourceforge.net/gazebo/xmlschema/#geom"
xmlns:joint="http://playerstage.sourceforge.net/gazebo/xmlschema/#joint"
xmlns:interface="
http://playerstage.sourceforge.net/gazebo/xmlschema/#interface"
xmlns:ui="http://playerstage.sourceforge.net/gazebo/xmlschema/#ui"
xmlns:rendering="
http://playerstage.sourceforge.net/gazebo/xmlschema/#rendering"
xmlns:controller="
http://playerstage.sourceforge.net/gazebo/xmlschema/#controller"
xmlns:physics="
http://playerstage.sourceforge.net/gazebo/xmlschema/#physics" >
<verbosity>5</verbosity>
<physics:ode>
<stepTime>0.005</stepTime>
<gravity>0 0 -9.8</gravity>
<cfm>0</cfm>
<erp>0.8</erp>
<quickStep>true</quickStep>
<quickStepIters>100</quickStepIters>
<quickStepW>1.3</quickStepW>
<contactMaxCorrectingVel>100.0</contactMaxCorrectingVel>
<contactSurfaceLayer>0.0</contactSurfaceLayer>
<updateRate>0</updateRate>
</physics:ode>
<rendering:gui>
<type>fltk</type>
<size>800 600</size>
<pos>0 0</pos>
</rendering:gui>
<rendering:ogre>
<ambient>0.8 0.8 0.8 1.0</ambient>
<sky>
<material>Gazebo/CloudySky</material>
</sky>
<grid>false</grid>
<shadowTechnique>stencilModulative</shadowTechnique>
</rendering:ogre>

<model:physical name="groundplane_model">
<xyz>0 0 0</xyz>
<rpy>0 0 0</rpy>
<static>true</static>
<body:plane name="ground_plane_body">
<geom:plane name="ground_plane_geom">
<normal>0 0 1</normal>
<size>1000 1000</size>
<segments>100 100</segments>
<uvTile>100 100</uvTile>
<material>Gazebo/GrassFloor</material>
<mu1>109999.0</mu1>
<mu2>1000.0</mu2>
</geom:plane>
</body:plane>
</model:physical>


<model:physical name="ball_model">
<xyz>0.025 0 1</xyz>
<static>false</static>
<body:sphere name="ball_body">
<geom:sphere name="ball_geom">
<xyz>0 0 0</xyz>
<size>0.015</size>
<mass>0.02</mass>
<visual>
<size>0.03 0.03 0.03</size>
<mesh>unit_sphere</mesh>
<material>Gazebo/White</material>
</visual>
<mu1>109999.0</mu1>
</geom:sphere>
</body:sphere>
</model:physical>


<model:renderable name="point_white">
<xyz>-3 0 5</xyz>
<static>true</static>
<light>
<type>point</type>
<diffuseColor>0.8 0.8 0.8</diffuseColor>
<specularColor>0.1 0.1 0.1</specularColor>
<range>20</range>

<attenuation>0.3 0.01 0</attenuation>
</light>
</model:renderable>

</gazebo:world>
-----------------------------------------------------------------------------------
Do you guys get the same behavior? If so, is there any fix or a way to get
rid of this?
Thanks in advance!
Rein

Post by Rein Appeldoorn
Hi,
I'm working on a simulation of an autonomous football table. When I model
the ball, it's sinking through the floor. Because the geometry of the

ball

------------------------------------------------------------------------------

------------------------------------------------------------------------------
EditLive Enterprise is the world's most technically advanced content
authoring tool. Experience the power of Track Changes, Inline Image
Editing and ensure content is compliant with Accessibility Checking.
http://p.sf.net/sfu/ephox-dev2dev
_______________________________________________
Playerstage-gazebo mailing list
https://lists.sourceforge.net/lists/listinfo/playerstage-gazebo

Rein Appeldoorn

2011-06-09 08:41:31 UTC

Permalink

Hi,

Thanks, when I play with kp and kd, I can prevent the ball from sinking. Now
another thing what is different from my ball model and yours:

Is it right that you set the parameters of a ball with a radius of 0.02 as
follows:

<size>0.02</size>
<visual>
<size>0.04 0.04 0.04</size>
</visual>

When I check this in my gazebo-simulation, it seems to be right. But in your
model you use: <scale> instead of <size>, is this right?

It's really weird because the examples use both (simpleshapes.world /
pioneer2dx.world).

Thanks,

Rein

here is a sample ball model I'm using for my simulations. Notice the

Post by Kamil Muszynski
<kd> and <kp> params and try to play with their values - they control
how much force
<body:sphere name = "ball_body">
<dampingFactor>0.0001</dampingFactor>
<xyz>0 0 0</xyz>
<rpy>0 0 0</rpy>
<geom:sphere name = "ball_geom">
<xyz>0 0 0.01</xyz>
<rpy>0 0 0 </rpy>
<size>0.02</size>
<mu1>0.3</mu1>
<kp>2e6</kp>
<kd>2e6</kd>
<bounce>120</bounce>
<bounceVel>0.5</bounceVel>
<slip1>0.7</slip1>
<slip2>0.7</slip2>
<mass>0.045</mass>
<visual>
<scale>0.02 0.02 0.02</scale>
<mesh>unit_sphere</mesh>
<material>Gazebo/Robocup_orange</material>
</visual>
</geom:sphere>
</body:sphere>
Also global parameters like step time, erp and cfm does matter, I'm
<stepTime>0.002</stepTime>
<erp>0.9</erp>
<cfm>0.01</cfm>
I hope that will help you,
Best regards,
Kamil Muszyński

Post by Rein Appeldoorn
I checked those two parameters but nothing changes. Is this a known fact

or
-----------------------------------------------------------------------------------

Post by Rein Appeldoorn
<?xml version="1.0"?>
<gazebo:world
xmlns:xi="http://www.w3.org/2001/XInclude"
xmlns:gazebo="http://playerstage.sourceforge.net/gazebo/xmlschema/#gz

Post by Rein Appeldoorn
xmlns:model="

http://playerstage.sourceforge.net/gazebo/xmlschema/#model"

Post by Rein Appeldoorn
xmlns:sensor="
http://playerstage.sourceforge.net/gazebo/xmlschema/#sensor"
xmlns:window="
http://playerstage.sourceforge.net/gazebo/xmlschema/#window"
xmlns:param="

http://playerstage.sourceforge.net/gazebo/xmlschema/#param"

Post by Rein Appeldoorn
xmlns:body="http://playerstage.sourceforge.net/gazebo/xmlschema/#body

Post by Rein Appeldoorn
xmlns:geom="http://playerstage.sourceforge.net/gazebo/xmlschema/#geom

Post by Rein Appeldoorn
xmlns:joint="

http://playerstage.sourceforge.net/gazebo/xmlschema/#joint"

Post by Rein Appeldoorn
xmlns:interface="
http://playerstage.sourceforge.net/gazebo/xmlschema/#interface"
xmlns:ui="http://playerstage.sourceforge.net/gazebo/xmlschema/#ui"
xmlns:rendering="
http://playerstage.sourceforge.net/gazebo/xmlschema/#rendering"
xmlns:controller="
http://playerstage.sourceforge.net/gazebo/xmlschema/#controller"
xmlns:physics="
http://playerstage.sourceforge.net/gazebo/xmlschema/#physics" >
<verbosity>5</verbosity>
<physics:ode>
<stepTime>0.005</stepTime>
<gravity>0 0 -9.8</gravity>
<cfm>0</cfm>
<erp>0.8</erp>
<quickStep>true</quickStep>
<quickStepIters>100</quickStepIters>
<quickStepW>1.3</quickStepW>
<contactMaxCorrectingVel>100.0</contactMaxCorrectingVel>
<contactSurfaceLayer>0.0</contactSurfaceLayer>
<updateRate>0</updateRate>
</physics:ode>
<rendering:gui>
<type>fltk</type>
<size>800 600</size>
<pos>0 0</pos>
</rendering:gui>
<rendering:ogre>
<ambient>0.8 0.8 0.8 1.0</ambient>
<sky>
<material>Gazebo/CloudySky</material>
</sky>
<grid>false</grid>
<shadowTechnique>stencilModulative</shadowTechnique>
</rendering:ogre>

<model:physical name="groundplane_model">
<xyz>0 0 0</xyz>
<rpy>0 0 0</rpy>
<static>true</static>
<body:plane name="ground_plane_body">
<geom:plane name="ground_plane_geom">
<normal>0 0 1</normal>
<size>1000 1000</size>
<segments>100 100</segments>
<uvTile>100 100</uvTile>
<material>Gazebo/GrassFloor</material>
<mu1>109999.0</mu1>
<mu2>1000.0</mu2>
</geom:plane>
</body:plane>
</model:physical>


<model:physical name="ball_model">
<xyz>0.025 0 1</xyz>
<static>false</static>
<body:sphere name="ball_body">
<geom:sphere name="ball_geom">
<xyz>0 0 0</xyz>
<size>0.015</size>
<mass>0.02</mass>
<visual>
<size>0.03 0.03 0.03</size>
<mesh>unit_sphere</mesh>
<material>Gazebo/White</material>
</visual>
<mu1>109999.0</mu1>
</geom:sphere>
</body:sphere>
</model:physical>


<model:renderable name="point_white">
<xyz>-3 0 5</xyz>
<static>true</static>
<light>
<type>point</type>
<diffuseColor>0.8 0.8 0.8</diffuseColor>
<specularColor>0.1 0.1 0.1</specularColor>
<range>20</range>

<attenuation>0.3 0.01 0</attenuation>
</light>
</model:renderable>

</gazebo:world>

-----------------------------------------------------------------------------------

Post by Rein Appeldoorn
Do you guys get the same behavior? If so, is there any fix or a way to

get

Post by Rein Appeldoorn
rid of this?
Thanks in advance!
Rein

Post by Rein Appeldoorn
Hi,
I'm working on a simulation of an autonomous football table. When I

model