tutorial inverse kinematic

[juerg]

Hi

After trying to follow the inverse kinematic turorial where all joints are in inverse kinematic mode I tried to build my on structure in a similar way - however - my joints get ignored in the simulation and everythings falls apart according to gravity (and manual).

I have not been able to understand why this is not happening with the tutorial robot?

Juerg

[coppelia]

Hello Juerg,

this can have several reasons. Probably your joints are not in the correct mode. Or your construction doesn't respect the required design rules (if using dynamics). It really depends on what you want to do.
The easiest would be to post your scene for us to have a look at it.

Cheers

[juerg]

Hi

I am most certainly happy to share my little example https://drive.google.com/folderview?id= ... sp=sharing

I might currently have quite some "funny" settings as I worked a while on it. My intention is to be able to move the "middle table" or the Sphere and record the "connectors" angels while doing this (something similar as the 7D Robot arm example is doing).

Currently the simulation gives me a stable situation (which is quite some advance to me from earlier designs)

Greetings

Juerg

[atoz]

Didn't check your files but if you haven't already, try putting your joints into IK mode and checking hybrid mode as well.

[coppelia]

Juerg,

in your scene, your object Basis is flagged as respondable. Why? What will actually happen is that object Basis and object ServoTraeger will collide and this produces the instability. You can also adjust the ndividual respondable masks to avoid those two object producing a collision response.

Cheers

[juerg]

hi
thanks for your input - I still can't see how to get a "ServoTable" that responds to my mouse movements.
I have set the Basis shape to not responding.
I have also set all revolute joins to inverse kinematic hybrid.
There will be motors in the implementation for the revolute joints magenta, blau and rot (the ones that are at the top of the table) and movements of these arms should result in a repositioning of the "table".

I have updated the scene (see above link)

Thanks, Juerg

[coppelia]

You have a dynamic construction, that is not using inverse kinematics. So you should set all your passive joints to force/torque mode, motors disabled. And all your active joints should be in force/torque mode, motors enabled, and position control enabled too.

Some of your joints are using a custom control. Don't use it, unless you know exactly what you are doing (usually PID control is good enough for many applications). You have to be careful, because you can have a lot of small bugs, that can be painful to find and correct.

Also, your mechanism would be overconstrained in the start position (the 3 active joints try to hold a horizontal position, which is not feasable for the given mechanism). Finally, I hope you noticed it, but your mechanism is not symmetric (shape weiss3 is smaller than the other equivalent shapes).

Cheers

[juerg]

hi
thanks - tried to look at my details and fix things.
could you please explain why you consider my construction overconstrained and what is causing the horizontal holding position?
I expected that by applying different forces to the motors the "table" should move away from the strongest arm?
Juerg

EDIT: I have adjusted the arm sizes to more reasonable values. Now the model (versuch3) shows much more a behaviour I expected? But I still can not see how to create a setup that will allow me to position the table with my mouse during simulation (as with the tutorial 7d robot arm).

[coppelia]

oh, I think I understand now what you want to do!
You basically want to position the table with the mouse, and have V-REP calculate the joint angles for the arms. Is that correct?

If that is what you want to achieve, you do not need to build a dynamic model. Following are the steps:

• prepare a IK group for the table. Basically your kinematic chain starts in the middle of the table, and goes over the sperical joint at the base. Make sure the spherical joint is in IK mode (do not check the hybrid mode box). Select a damped resolution method, and constrain the IK group in the x,y and z axis.
• prepare 3 IK groups, one for each arm. Each IK group has the tip where the spherical joints are currently located. And the base on the table. Make sure that all joints are simply in IK mode (do not check the hybrid mode box). Select a non-damped resolution method, and constrain the IK group in the x,y and z axis (remove the spherical joints at the tip)
• make sure that all shapes are static and non-respondable

I believe that you mechanical construction still needs additional joints. At least one revolute joint around the vertical axis of the table (in V-REP, spherical joints give you 2 DoFs, not 3)

Cheers

[juerg]

Hi
Real progress now and Mark Freese has already followed your instructions and sent me an updated version.
I hope I am not exausting your patience - after setting all IK-Groups to DLS I have now a stable simulation - however I can't move the table - or do I have a mouse control problem?

https://drive.google.com/folderview?id= ... sp=sharing

versuch4.tts

Juerg

P.S. the table should only move in X/Y so I think spherical at the table base is what I want