Possible wrong position of waist IMU ergoCubSN001

[fils99]

Task description

Today, during experiments with ergoCubSN001, we wanted to check the position and orientation of the waist IMU. First of all we looked to the IMU in the real robot. Afterwards, we compared them with the ones of the URDF model (we use this version of ergocub-software e4b567e).

As we can see in the image, the position of waist IMU in the real robot seems a bit different wrt the URDF one.

Definition of Done

Understand if position of waist IMU is correct

cc @LoreMoretti @GiulioRomualdi

[GiulioRomualdi]

cc @Nicogene and @traversaro

[Nicogene]

The imu in the waist has been added in the SN000 urdf as XML blob (since the simulation model cad is unmaintained):

• Add waist_imu in urdf file #99

I didn't recall that is still exported as XML blob also in newer models:

ergocub-software/urdf/creo2urdf/data/ergocub1_1/base_1_1.yaml

Lines 1142 to 1150 in b030a60

	- |
	<link name="waist_imu_0"/>
	- |
	<joint name="waist_imu_0_fixed_joint" type="fixed">
	<origin xyz="0.0354497 0.00639688 0.060600" rpy="1.5708 0 -1.5708"/>
	<parent link="torso_1"/>
	<child link="waist_imu_0"/>
	<dynamics damping="0.1"/>
	</joint>

Maybe @Mick3Lozzo recall why we didn't exported that frame, but we should do it from the cad (cc @fiorisi @Lawproto)

[fiorisi]

If I correctly understood the problem I think that the IMU was moved because of a possible collision with the torso covers: https://github.com/icub-tech-iit/project-ergocub/issues/107#issuecomment-1531510392.

[Nicogene]

If I correctly understood the problem I think that the IMU was moved because of a possible collision with the torso covers: icub-tech-iit/project-ergocub#107 (comment).

Good catch!

[Nicogene]

Today I checked the transform between the SCSYS_TORSO_1 and the waist IMU csys and here is the result

I used this snippet using the idyntree binding for getting xyzrpy:

rpy = idyntree.bindings.Rotation(0, 0, -1, -1, 0.000105208, 0, 0.000105208, 1, 0).asRPY()

And I got:

-<origin xyz="0.0354497 0.00639688 0.060600" rpy="1.5708 0 -1.5708"/> 
+<origin xyz="0.0606 -0.00639428 0.01745" rpy="1.5708 -0.000105208 -1.5708"/> 

The difference seems not only a displacement, I used the correct CSYS for computing the transform? Are those csys correct? I took them from ecub.asm (cc @fiorisi @Mick3Lozzo)

[Nicogene]

xref:

• ergoCub SN001/ ergoCubGazeboV1_1*: the waist imu is still added as xml blob to the urdf #215

[Mick3Lozzo]

xref:

• ergoCub SN001/ ergoCubGazeboV1_1*: the waist imu is still added as xml blob to the urdf #215

@Nicogene what kind of IMU support is AMI using? I am going to check.

cc @fiorisi

[Nicogene]

xref:

• ergoCub SN001/ ergoCubGazeboV1_1*: the waist imu is still added as xml blob to the urdf #215

@Nicogene what kind of IMU support is AMI using? I am going to check.

cc @fiorisi

The xsens mti that is mounted in the torso, I got that transform pulling the latest cad-mechanics

[Mick3Lozzo]

The difference seems not only a displacement, I used the correct CSYS for computing the transform. Are those csys correct? I took them from ecub.asm (cc @fiorisi @Mick3Lozzo)

These are the same values I found @Nicogene.

[Nicogene]

Thanks @Mick3Lozzo !
That very small rotation on pitch is strange ( we can also approximate it to zero ), but if this is the correct pose I will change the urdf accordingly

[Nicogene]

Something has to be fixed:

• Fix the position of the imu in the waist #267 (comment)

Probably on creo2urdf side

[Nicogene]

Hi @fils99, with:

• Fix the position of the imu in the waist #267

The position of waist_imu_0 is:

That seems coherent with

[fils99]

Thanks @Nicogene!