Skip to content
This repository has been archived by the owner on Jul 23, 2020. It is now read-only.

Tactile clean #233

Open
wants to merge 11 commits into
base: master
Choose a base branch
from
Open

Tactile clean #233

wants to merge 11 commits into from

Conversation

rafaelxero
Copy link
Contributor

Simulates a tactile sensor by providing extracted contact forces.

@s-nakaoka
Copy link
Owner

Let me point out the followings:

  • OpenRTMPlugin was separated from the Choreonoid source repository, and it is now an independent module called choreonoid-openrtm. ( https://discourse.choreonoid.org/t/openrtm/363 ). You have to modify the source code to accommodate the latest configuration.

  • I need more detailed information on the plugin to determine whether to merge it. Could you provide such information?

@rafaelxero
Copy link
Contributor Author

I am sorry for my very late reply. I was going to write earlier but then got busy and forgot to reply. Sorry for that.

After submitting the Pull Request I found that there were some improvements that needed to be done to the code and now there are still some bugs that I still have to fix, besides modifying it to account for the fact that OpenRTMPlugin is now independent.

Can we have this Pull Request pending? or would you prefer me to close it and re-open it later once I have fixed the bugs?

In the following I provide more detailed information:

The simulated tactile sensor of the Pull Request requires the plugin ContactForceExtraction
The idea is to collect the forces that are applied to a certain surface of a link and perform a spatial discretization of them according to a grid of cells. The parameters of the grid are defined in the VRML of the robot:

  • translation and rotation of the plane of the sensor
  • size of the rectangle (minX, maxX, minY, maxY) in the sensor frame
  • number of rows and cols of the matrix

https://github.com/isri-aist/hrp5p/blob/d2c810e004f06d1a77478bb28f0547d2b45a20ea/model/HRP5Pmain_TS-SMX1604.wrl#L477

The discretization works as follows:

  • All the forces that lie within on cell are summed and placed at the center of the cell
  • If the force is outside of the rectangle, it is saturated to the boundary cells

https://github.com/rafaelxero/choreonoid/blob/4733c688ea9eb3c65869b96f78a4ba07a625d46c/src/Body/ForwardDynamicsABM.cpp#L526

The tactile information is published through the RTC of the robot:

https://github.com/isri-aist/hrpcnoid/blob/17ac705185ecf7f11f378abb7125042d1946ab52/model/HRP5P/Virtual-HRP5P-tactile-RTC.conf#L22

This information tries to resemble the data provided by the tactile sensors that we recently developed for HRP-5P and which are implemented as FSR matrices.
Although the simulated tactile information only provide outer points (no inner points) we are using the convex hull of the forces. So, we still can use the simulated tactile information to assess its use to walk over surfaces that are narrower than the feet.

@s-nakaoka
Copy link
Owner

I'm sorry for my late replay, too.
Thank you for describing the details about what you are developing.

Can we have this Pull Request pending? or would you prefer me to close it and re-open it later once I have fixed the bugs?

OK, I will put this pull request on hold.

Please note that you have to improve the code so that it can work with OpenRTMPlugin, which is the external plugin now.
Also, I'm grad if you can separate your functions from the main Choreonoid source as a plugin implementing the functions. That makes it easier to maintain the source code.

Thank you for your cooperation in advance.

@rafaelxero
Copy link
Contributor Author

Thank you for waiting...
Recently there have been other projects that have prevented me to restart the development of this issue.
As soon as I work on this again I will let you know. Sorry.

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants