## How to accurately link a parallel configuration with dummy?

Typically: "How do I... ", "How can I... " questions
gloria7@scuthahaha
Posts: 10
Joined: 15 Oct 2017, 10:01

### How to accurately link a parallel configuration with dummy?

Hello:

I'm figuring out a parallel configuration as a shoulder part of a robotic arm. And I read that the v-rep contain the inverse kinematics modules, just export the STL model into the v-rep scene and fix the physical structure and assembly tree correctly, then add the dummy into the end of the tree and you can get the model works well.
But things didn't work as I put the inverse tip-target into my model, and the model link is :https://drive.google.com/open?id=1rsK4A ... tMI83okiLi.
There are 3 questions:
1, As my model is a 3-DoF configuration, only the under part of gear can moving passively and other joint should run as the restraint of the model, but as I run the simulation, it works only to the gear part and other joint didn't run.
2, How to assemble the parallel configuration? I choose to make the upper platform under the gear3 chain and use tip-target to restraint the model but it didn't works well.
3, My system is macOS so I cannot use the vortex or video record tools in 3.5.0 version, is there any resolution to solve it?(but it actually not essential for I can use other applications to record the scene: >)

Thank you,
Gloria

coppelia
Posts: 7396
Joined: 14 Dec 2012, 00:25

### Re: How to accurately link a parallel configuration with dummy?

Hello Gloria,

the link you provided is not public...

Cheers

gloria7@scuthahaha
Posts: 10
Joined: 15 Oct 2017, 10:01

### Re: How to accurately link a parallel configuration with dummy?

Hello：

cheers,
Gloria

coppelia
Posts: 7396
Joined: 14 Dec 2012, 00:25

### Re: How to accurately link a parallel configuration with dummy?

You have to decide what kind of simulation you want to achieve: kinematic or dynamic. Be aware of the differences. Basically, with kinematics you'll achieve precise and exact calculation, while with dynamics, things will be slower, more wobbly and less precise, since things are resolved differently and by the physics engine (i.e. masses and inertias need to be taken into account, etc.).
I'd really go with a kinematics simulation. Have a look at this tutorial, and also the demo model Models/robots/non-mobile/ABB IRB 360.ttm for example.

Currently your model appears to be a kinematic model, but there are a few dynamically enabled parts which doesn't make sense.

Cheers

gloria7@scuthahaha
Posts: 10
Joined: 15 Oct 2017, 10:01

### Re: How to accurately link a parallel configuration with dummy?

Hi, coppelia:

your resolution is very useful and I have solved the question of kinematic model I think.But there is an other question that as I reference to the model of irb360 in vrep, I read the control code of it and migrated it to my model. I didn't understand the usage of

Code: Select all

sim.rmlMoveToJointPositions(fkDrivingJoints,-1,currentAngVel,currentAngAccel,maxAngVel,maxAngAccel,maxAngJerk,{-80*math.pi/180,0*math.pi/180,0*math.pi/180},targetAngVel)
therefore, my model perform like this:
like the control code has effect on dummy but not my joint? And the inverse kinematic of my model still don't works.That's confused me so much.
Hope you can give me some instructions, many thanks.

Yours,
Gloria

coppelia
Posts: 7396
Joined: 14 Dec 2012, 00:25

### Re: How to accurately link a parallel configuration with dummy?

So you have decided for a kinematic simulation. Now you need to decide whether you want IK, FK, or both.

With your mechanism, IK alone would be the simplest to implement (simply 3 kinematic branches that close at the end-effector or upper platform). FK is more complex since you will need a branched kinematic, a Y-type tree where the tree base comes from the bottom, and the two tree branches close at the bottom.
IK/FK is even more complex.
I highly recommend that you have a close look at how the demo model Models/robots/non-mobile/ABB IRB 360.ttm was made.

Cheers

gloria7@scuthahaha
Posts: 10
Joined: 15 Oct 2017, 10:01

### Re: How to accurately link a parallel configuration with dummy?

Hi Coppelia:
Yes, I want to develop a kinematic simulation. Thanks for your reply so I decided to do the inverse solution first. But the question is that when I start the simulation ,it will report the error that :
Lua runtime error: [string "CHILD SCRIPT target1"]:38: Object not tagged for explicit handling. (sim.handleIkGroup)
stack traceback:
[C]: in function 'handleIkGroup'
[string "CHILD SCRIPT target1"]:38: in function <[string "CHILD SCRIPT target1"]:15>
So I think the question is my ik-tip group I set was wrong. (but I have already look at the model of irb360 and learn the link method of it)And my supervisor suggest me to write the Jacobin matrix by myself and do the inverse kinematic model on my own. So is there any example to write the inverse kinematic script?

Yours,
Gloria

coppelia
Posts: 7396
Joined: 14 Dec 2012, 00:25

### Re: How to accurately link a parallel configuration with dummy?

Did you check out this tutorial?

Cheers

gloria7@scuthahaha
Posts: 10
Joined: 15 Oct 2017, 10:01

### Re: How to accurately link a parallel configuration with dummy?

Hi Coppelia：

Thanks for your reply. But my model still not work out :(

I think the "Y" types link method is good, so the tree is like this:

I have already check the inverse kinematic model of this parallel configuration and used the right setting of tip-target link. So I can realize the inverse kinematic model. But My project is how to simulate the singularity configuration and do the motion planning of the configuration(avoid the singularity).Thus, now I want to do both inverse and forward kinematic of the model.

I have check the irb360 model in the models. It inspired me much, It should contend a XYZ cartesian to give the position of upper platform .And I have read the code in it many times.

Here are some question of irb360 models, which I think if I could figure out can make many helps to my own models.

1. I didn't known which parameter controls the driving joint(passive mode).At first I think the lines of setFk or setIk are corresponding to the number of initial code but if seems didn't work like this way.
2. I have noticed that the irb360 has some branch with joints only, I think it was for the link to other model use.

And back to my model, of course it was very simple, only 3 links with 2 platform. As I have move the irb360 script to my model, it didn't works and return "ik solver didn't works"error. But the Fk seems works, for the model can rotate. However, the rotate method is not what I want

1, only link 3 's gear3joint1 moved as the script execute, so it represent that the motion of the structure is not change. And I want do the analysis of singularity of the parallel configuration. So it was not what I want.
2, as the method the script used was to gave the degree of angle that the passive joint moved , and I want the degree can be control with some GUI like slider? So that I can control the degree of angle that I want the passive joint moved.
3, if the inverse kinematic of my model doesn't works, is that means my configuration cannot be simulate with the Ik tools, and I should right the inverse kinematic own(like write the Jacobin matrix of this model?)

I should solve these ASAP, for my graduate design has limited times, and if it doesn't works I cannot write my graduate paper :( As my senior cannot give me any advice of the model with vrep, and I firmly believe that it can do the parallel configuration simulation, I feel very anxious and fear not graduate.

Hope you can give me some tips to solve with my project.

And here is my latest modelhttps://pan.baidu.com/s/1kmrCmzcjAIvqAmPcny3hWQ password: ttha
(sorry I can't use google now)

Thanks,
Gloria
2018.4.24

gloria7@scuthahaha
Posts: 10
Joined: 15 Oct 2017, 10:01

### Re: How to accurately link a parallel configuration with dummy?

Hello coppelia:

I have noticed my problems of the PKM. As the geometrical feature of the Gosselin structure, it has only 3 DOF.I compared the model with Irb360 and found that it lacks many joints which contain in the delta structure.The irb360 models used 3 revolute joint to replace the spherical joint. And it has a Forward kinematic tree to calculate the forward kinematics(as you can see the one branch of tree was build up with joint only and I think its for the calculation used)
But my structure has many interesting features, like it center of upper platform always on the sphere which sphere center is located at ground platform circle's center.

As it has these strict properties, I wonder if the inverse kinematic in vrep cannot calculate it well, so I want to rewrite the kinematic of it.But the matrix of Lua is limited(eg. cannot do the diag or det calculation ).And I don't know how to control tip-dummy with external kinematic.
Forgive my limited expression of academic English.Our project has already 3D print the model of the PKM. What I want to do is simulate the PKM in vrep and analysis the singularity of the model. I record a video of our model to you so you can understand my question better.