Review and Unify Joint Directions of Rotation

[jgvictores]

Review Joint Directions of Rotation https://github.com/roboticslab-uc3m/teo-developer-manual/blob/master/appendix/a-teo-diagrams.md#joint-directions-of-rotation then unify

[rsantos88]

I found that some directions of rotation are wrong.
Do you want to correct the diagram or the direction in the real robot?

[rsantos88]

The signs are ok but, @jmgarciah and me think that the arrows can generate confusin in the Y axis

[jmgarciah]

Talking with @smcdiaz and seeing the TEO's hardware, we have found a possible solution. Related with the two saggital motor of the legs, the wires of the gear are different from the rest.

@jgvictores have given me permission to change the order of the wires and have the same turning sense that the diagram.

[jmgarciah]

I have found an image that could be insteresting for a new TEO diagram. And finally understand the joint direction.

[jgvictores]

Looks nice, could be done with TEO.

I would try to avoid the RPY terminology (Roll is Z in some fields) and stick to ours.

[jmgarciah]

I totally agree. It is only an idea. So please, take it as you need

[RaulFdzbis]

In my opinion, i think that overlapping of the robot model over the axis, make that image a little overload, and can be confusing.

Maybe a good solution could be to use the simplified model that is used now with the Teo Joint Directions of Rotation but with the perspective in that image, for an easier understanding.

[jgvictores]

If anyone could take a picture of TEO in a pose similar to NAO there (same perspective, but arms pointing down), I could give it a shot. Some alpha layer magic should do the trick. :-)

[David-Estevez]

This is not a picture, but might be useful:

[jmgarciah]

Something like @David-Estevez shows could be perfect

[David-Estevez]

I can pose it and change the colors to whatever suits your diagrams, just tell me!

[rsantos88]

@David-Estevez totally agree 👍

[jmgarciah]

@smcdiaz and me have finished to correct the saggital joint direction. But we have to check that everything is ok

[rsantos88]

@smcdiaz and me have finished to correct the saggital joint direction. But we have to check that everything is ok

I've checked the result with @jmgarciah and it's not working well.
When I run launchLocomotion and check the value of sagital axis, the value is -3.93673706054688
I move this axis to homing position, setting position to 0, and the axis moves outside and when I get the position, it returns -7.45166015625
@jmgarciah propouse to @jgvictores or @smcdiaz to have a look at this problem because
he has no idea.

[rsantos88]

Changed sign of CUI Absolute Encoder to read the correct joint directions of rotation.
See here the changes. It has been tested and it works ok. Now the directions of rotation of the joint ID 5 and 12 correspond to the diagram.

Note: the axial rotation of the waist has not been corrected

[rsantos88]

notes:

• PR is pending to be performed, but it has been changed on the robot to be use.
• After all the changes, I've corrected the home position of the legs (more accurate initial position)

[rsantos88]

Other curious note observed:

• When you run launchLocomotion, always the right sagittal leg moves arround 10º degrees out.
  You can correct this doing the home position with --homePoss, but something causes the joint to move when the driver is initializing.

[jmgarciah]

The legs' sagittal joints are already fixed. The only remaining joint is the axial waist. Fixing via hardware this joint is more complicate than via software, I think. It is neccesary to separe the body from the legs. I propose to change the values from .ini files. Please @jgvictores , help us.

[jgvictores]

On my way!

[jgvictores]

It is neccesary to separe the body from the legs. I propose to change the values from .ini files

First diagrams, then update all related configuration files and robot models.

[jgvictores]

The fact that there are negative trs in launchLocomotion.ini means that diagrams are incorrect.
I've opened roboticslab-uc3m/teo-configuration-files#11 pointing at the situation.

[rsantos88]

The fact that there are negative trs in launchLocomotion.ini means that diagrams are incorrect.
I've opened roboticslab-uc3m/teo-configuration-files#11 pointing at the situation.

roboticslab-uc3m/teo-configuration-files#11 has been closed.

[rsantos88]

The direction of rotation of the axial waist is opposite in the .ini config file and the diagram.
What we should to change? The diagram? The config file? or it doesn't matter which of them?

[jgvictores]

@rsantos88 I've taken a look at roboticslab-uc3m/teo-configuration-files@5addf18 and it seems amazing!

• To close this issue I'd be satisfied with real TEO's ini files and manual images, and open separate issues on simulation model repositories
• Note to self: develop small script to enable porting programs to be compatible with new criteria

Issues that will be unblocked:

• Joint limits: see my comment at Review and Unify TEO model joint limits #15 (comment)
• Kinematic model and diagrams: Review and Unify TEO kinematic model #38
• Probably many more!

[rsantos88]

To close this issue I'd be satisfied with real TEO's ini files and manual images, and open separate issues on simulation model repositories

I've modified the images of directions. Review it please: roboticslab-uc3m/teo-developer-manual@6238c4b

In my humble opinion, I think it would be quite interesting to reconsider the revision of the limits. In the arms a more detailed revision of the limits has already been made, correcting some errors. (#15 (comment))
I would work on these limits. Also, I would recommend checking the limits of the legs again with @jmgarciah , and apply the new changes of direction. I await your proposals :)

To continue with this issue, once again, I would recommend modifying and correcting the limits of the arms and legs and then the direction of rotation, are you agree with that?

[jgvictores]

Looks great to me!

IMHO, you can close the issue, and we can proceed with:

• Review and Unify TEO model joint limits #15 (joint limits)
• Update simulated models (or put issue per repo)
• Update demos (or put issue per repo)
  • https://github.com/roboticslab-uc3m/waiter
  • https://github.com/roboticslab-uc3m/follow-me
  • https://github.com/roboticslab-uc3m/teo-self-presentation

[rsantos88]

Here are the result of merging the new joint limits of arms with the new direction of rotation in "teo-configuration-files" repository: roboticslab-uc3m/teo-configuration-files#18
(please, check it in case something has escaped me)

Maybe we can keep it this way, without revising the limits of the legs again and consider that the previous measurement with @AlvaroMartinezR was correct. What do you think? Would it be interesting to review the limits of the legs again before continuing?
If we continue without legs revision, the next steps would be:

• Reflect the changes in the tables of teo-developer-manual repository
  • Accept pull request: New direction of rotations, joint limits and updated human inspired limits teo-developer-manual#50
• Finish the changes in the .ini configuration files:
  • Accept this pull request: Merging new direction of rotation and joint limits teo-configuration-files#18
• Apply changes in yarp-devices repository (this has been tested in POS and VEL mode.)
  • Accept pull request: Applied absolute value for tr in setMinLimitRaw and setMaxLimitRaw yarp-devices#205

and...

• Update simulated models (or put issue per repo)
  • Accept this pull request: Updated joint limits and direction of rotations in Teo Openrave Model teo-openrave-models#25
• Update demos (or put issue per repo)
  • https://github.com/roboticslab-uc3m/waiter
  • https://github.com/roboticslab-uc3m/follow-me
  • https://github.com/roboticslab-uc3m/teo-self-presentation

[jmgarciah]

I think you should check the whole robot, not just the arms. If we are going to modify the directions of the joints, we should not leave the work halfway.

[rsantos88]

Updated: #39 (comment)

[jgvictores]

• Accept pull request (trs in yarp-devices): roboticslab-uc3m/yarp-devices#205

Merged at roboticslab-uc3m/yarp-devices@304b245

[jgvictores]

• Accept pull request: roboticslab-uc3m/teo-developer-manual#50

Merged at roboticslab-uc3m/teo-developer-manual@118f472

[jgvictores]

• Accept this pull request: roboticslab-uc3m/teo-configuration-files#18

Merged at roboticslab-uc3m/teo-configuration-files@23703e1

[jgvictores]

[x] Update demos (or put issue per repo)

• https://github.com/roboticslab-uc3m/waiter
• https://github.com/roboticslab-uc3m/follow-me
• https://github.com/roboticslab-uc3m/teo-self-presentation

Will be tracked at:

• Update joint-space commands teo-waiter#9
• Update joint-space commands teo-follow-me#46
• Update joint-space commands teo-self-presentation#8

[rsantos88]

Updated #39 (comment) about update simulated models

[jgvictores]

[x] Update simulated models (or put issue per repo)

• Update joint directions and limits teo-openrave-models#26 (Updated joint limits and direction of rotations in Teo Openrave Model teo-openrave-models#25)
• Update joint directions and limits teo-simox-models#3
• Update joint directions and limits teo-ros#1
• Unify robot GAZEBO model joint directions and limits legacy-teo-gazebo#5

[rsantos88]

Ok! Can I close this issue or do you want to add more things like the change in kinematics (important issue)?

[jgvictores]

Yes, please close, and we can actually unblock kinematics! Other issues are all tracked individually. 🎉🎉🎉