ergoCub 1.1 S/N:001 – Fingers' positions are too much noisy

[PasMarra]

Robot Name 🤖

ergoCub 1.1 S/N:001

Request/Failure description

Even when the fingers are kept still, their measured positions is varying too much over time because of the noisy measurements.

Detailed context

Here you can find some useful logging file: https://istitutoitalianotecnologia-my.sharepoint.com/:f:/g/personal/pasquale_marra_iit_it/EiYZm4pUaRBMlSTIA5OfwaYBPdCQ2MF5ZT_LZ-7tUBccHQ?e=Duk9kV

Additional context

The fingers on the real robot are not moving, yet this is what happens:

fingers_trembling.webm

How does it affect you?

No response

[Nicogene]

@PasMarra a curiosity, that visualization is rviz?
If yes, how are you publishing the decoupled fingers joints?

[PasMarra]

Hi @Nicogene, I've created a module to read the joint states from the /joint_states topic and, given the parameters and the coupling laws, I evaluate and publish a new message on a new topic (/joint_states_full), from which the robot_state_publisher can retrieve the full state.

[github-actions]

This issue has been automatically marked as stale because it did not have recent activity. It will be closed if no further activity occurs.

[Nicogene]

Hi @PasMarra!

we are aware that the measures of the fingers encoders may be affected by noise, it would be very useful if you can post a plot of the fingers encoder readings in order to understand the magnitude of the problem.

,I've created a module to read the joint states from the /joint_states topic and topic and, given the parameters and the coupling laws, I evaluate and publish a new message on a new topic (/joint_states_full

Moreover, if you point us to the software you are using (also privately in case it is not sharable), this would help debug the problem.

Thanks!

cc @maggia80

[PasMarra]

Hi @Nicogene,
unfortunately I will not be back at CRIS earlier than April 8th and I've noticed that the attached rosbag

Here you can find some useful logging file: https://istitutoitalianotecnologia-my.sharepoint.com/:f:/g/personal/pasquale_marra_iit_it/EiYZm4pUaRBMlSTIA5OfwaYBPdCQ2MF5ZT_LZ-7tUBccHQ?e=Duk9kV

does not contain any encoders' measurements, only the /joint_states and '/joint_states_full' topics, so I cannot provide any of the plots you are asking for.

I'm gonna send you privately the software I'm using to publish the full joints state.

[Nicogene]

/remind 8th April Revive this activity

[octo-reminder]

⏰ Reminder
Monday, April 8, 2024 10:00 AM (GMT+02:00)

Revive this activity

[octo-reminder]

🔔 @Nicogene

Revive this activity

[PasMarra]

Hi @Nicogene, these are the plots for each encoder when the fingers are kept still.

Notice that the l_thumb_add and r_thumb_oc signals are completely flat, maybe thay are not working at all.

LEFT HAND

Left hand

LEFT FINGERS

l_thumb_oc

l_thumb_add

l_index_oc

l_index_add

l_middle_oc

l_ring_pinky_oc

RIGHT HAND

Right hand

RIGHT FINGERS

r_thumb_oc

r_thumb_add

r_index_oc

r_index_add

r_middle_oc

r_ring_pinky_oc

[Nicogene]

Thanks @PasMarra ! Are those plots in rad ?

[PasMarra]

Thanks @PasMarra ! Are those plots in rad ?

Yes, they are.

[maggia80]

The TLV493D has this accuracy in the non-calibrated mode (we are using non-calibrated mode):

taken from: https://www.infineon.com/dgdl/Infineon-3D_Magnetic_Sensor_for_Angle_Measurements-AN-v01_10-ApplicationNotes-v01_01-EN.pdf?fileId=5546d46265f064ff01665a3d22f055e3

cc: @marcoaccame

[maggia80]

the magnet used is this one: https://www.magfine.it/it/products/detail.php?product_id=22298

[lornat75]

would you expect the error reported in your plot to correspond to variable noise as pasquale reported? it is weird that some encoders/fingers have different level of noise.

[maggia80]

Channels with more noise could be related to the distance of the magnet from the sensor (we should check this correlation).
It is worth investigating and seeing if we can reduce this noise.

[simeonedussoni]

hi @PasMarra
given that the forearms have been reworked, this effect is still present? can we have a hands-on with the robot to investigate the origin of noise?

[PasMarra]

hi @PasMarra given that the forearms have been reworked, this effect is still present?

Hi @simeonedussoni, I haven't had the opportunity to check it.

can we have a hands-on with the robot to investigate the origin of noise?

Yes, sure. Let's arrange a lab session on teams.