make a dynamic simulation for Stewart Platform

Requests or suggestions for new features
biscuit122
Posts: 6
Joined: 23 Jan 2018, 16:16

make a dynamic simulation for Stewart Platform

Post by biscuit122 » 26 Jan 2018, 14:48

I have two problems:
1. how to build a cardan joint using two revolute joints to accomplish the dynamic simulation?
I have tried several times to use a sphere pure shape to connect the two revolute joints, and add two dummy to the sphere pure shape and other shape. But the cardan joint I made cannot support the platform.

before simulationImage

start to simulate Image

2. how to build a joint with 2 DOF of transformation and revolution, like a electrical cylinder?

My modelImage

coppelia
Site Admin
Posts: 6511
Joined: 14 Dec 2012, 00:25

Re: make a dynamic simulation for Stewart Platform

Post by coppelia » 27 Jan 2018, 16:03

Hello,

I suspect that your joint is generating internal collisions (enable contact point display to verify that). The other suspicion is that your masses and inertias might be wrong. Make sure to carefully read this page, mainly the design considerations 7 and 8.

Cheers

biscuit122
Posts: 6
Joined: 23 Jan 2018, 16:16

Re: make a dynamic simulation for Stewart Platform

Post by biscuit122 » 28 Jan 2018, 09:10

coppelia wrote:
27 Jan 2018, 16:03
Hello,

I suspect that your joint is generating internal collisions (enable contact point display to verify that). The other suspicion is that your masses and inertias might be wrong. Make sure to carefully read this page, mainly the design considerations 7 and 8.

Cheers
Thanks for your help, I have solved my problems. But what is the influence result from this modifying approach? For example, the force supported by electrical cylinder to upper platform?

Thanks very much!

coppelia
Site Admin
Posts: 6511
Joined: 14 Dec 2012, 00:25

Re: make a dynamic simulation for Stewart Platform

Post by coppelia » 29 Jan 2018, 20:30

in general, try to keep the overall mass same, and even-out the masses internally, in order to avoid that kind of simuations. Same for the inertias. Those requirements are not that important with the Vortex engine (and to a certain extent with the Newton engine). So you should always switch engines to see if you obtain very different simulation behaviours.

Cheers

biscuit122
Posts: 6
Joined: 23 Jan 2018, 16:16

Re: make a dynamic simulation for Stewart Platform

Post by biscuit122 » 30 Jan 2018, 04:56

coppelia wrote:
29 Jan 2018, 20:30
in general, try to keep the overall mass same, and even-out the masses internally, in order to avoid that kind of simuations. Same for the inertias. Those requirements are not that important with the Vortex engine (and to a certain extent with the Newton engine). So you should always switch engines to see if you obtain very different simulation behaviours.

Cheers


Hello, even though I have solve the joint problem using two serial joints to form one joint. I found my model doesn't fall, but the cardan joints cannot operate. I set 6 prismatic joints with same target position 0.1m and the rest joints to passive-hybrid mode. The prismatic joints cannot arrive in its target position. I think it is resulted from the locked cardan joints I made. But I don't know how to set it to operate normally.

This is the my problematic model.

Thank you very much.
Last edited by biscuit122 on 30 Jan 2018, 15:55, edited 2 times in total.

coppelia
Site Admin
Posts: 6511
Joined: 14 Dec 2012, 00:25

Re: make a dynamic simulation for Stewart Platform

Post by coppelia » 30 Jan 2018, 08:50

any particular reason you perform a dynamic simulation? Do you need to measure forces in the platform joints?
You will never get as stable and precise results using the dynamics engine for your calculation. If possible, make the kinematic representation of your platform, and set the platform end-effector as static, respondable shape: this way you can still dynamically manipulate items.

Is it possible that your joints simply don't have enough range? It seems their limits is between -0.15 and 0. If the joint target position is specified as 0.1 (as in your case), you won't be able to reach it.

Cheers

biscuit122
Posts: 6
Joined: 23 Jan 2018, 16:16

Re: make a dynamic simulation for Stewart Platform

Post by biscuit122 » 30 Jan 2018, 16:11

coppelia wrote:
30 Jan 2018, 08:50
any particular reason you perform a dynamic simulation? Do you need to measure forces in the platform joints?
You will never get as stable and precise results using the dynamics engine for your calculation. If possible, make the kinematic representation of your platform, and set the platform end-effector as static, respondable shape: this way you can still dynamically manipulate items.

Is it possible that your joints simply don't have enough range? It seems their limits is between -0.15 and 0. If the joint target position is specified as 0.1 (as in your case), you won't be able to reach it.

Cheers
Yes, I want to measure forces of supporting upper platform from cardan joints. Is the end-effector you mentioned the upper platform or base platform? Do you mean that I cannot achieve my simulation with dynamic model? Actually, this is just one leg of our robot. Our quadruped robot has 4 paralell Stewart legs in Adams. Now we want to accomplish this simulating model in vrep. Can we achieve that?


And the prismatic joint has no problem in my opinion. Their limits are betwwen -0.15 and 0.15. I think this is resulting from the cardan joints, which cannot rotate flexibly.

coppelia
Site Admin
Posts: 6511
Joined: 14 Dec 2012, 00:25

Re: make a dynamic simulation for Stewart Platform

Post by coppelia » 30 Jan 2018, 18:21

Your prismatic joints have a problem: their position minimum is -0.15, and their range is 0.15. (range is not the upper limit). If you want an upper limit of 0.15, then you should specify a range of 0.3

You will probably have to tweak your simulation quite a bit to make your 4-legged robot walk with the 4 stewart platforms. In any case you should try using the Vortex engine, which is the most robust among all 4 engines supported.

Cheers

biscuit122
Posts: 6
Joined: 23 Jan 2018, 16:16

Re: make a dynamic simulation for Stewart Platform

Post by biscuit122 » 31 Jan 2018, 08:21

coppelia wrote:
30 Jan 2018, 18:21
Your prismatic joints have a problem: their position minimum is -0.15, and their range is 0.15. (range is not the upper limit). If you want an upper limit of 0.15, then you should specify a range of 0.3

You will probably have to tweak your simulation quite a bit to make your 4-legged robot walk with the 4 stewart platforms. In any case you should try using the Vortex engine, which is the most robust among all 4 engines supported.

Cheers

Thanks for your suggestions. I have solved mentioned problems.

Because of unfamilirity of vrep, I have made 2 kinds of model to achieve same action.

http://pic.suiyiyun.cn/597454/6DOFStewart1-1.ttt
Model 1 used a spherical joint as cardan joint, and only one prismatic joint to achieve motion of electrical cylinder.

http://pic.suiyiyun.cn/597454/6DOFStewart1-4.ttt
Model 2 used two revolute joints as cardan joint, and prismatic joint and revolute joint to achieve motion of electrical cylinder.

Now I control them using a same Matlab program via remote API. When I set a displacement along X-axis (X-axis and Y-axis point to right and forward direction), I found that Model 2 happened extra shake like this.http://pic.suiyiyun.cn/597454/model2.avi

The normal operation of Model 1 is shown in http://pic.suiyiyun.cn/597454/model1.avi



Do you know the reason?


Thanks a lot.

biscuit122
Posts: 6
Joined: 23 Jan 2018, 16:16

Re: make a dynamic simulation for Stewart Platform

Post by biscuit122 » 31 Jan 2018, 13:58

coppelia wrote:
30 Jan 2018, 18:21
Your prismatic joints have a problem: their position minimum is -0.15, and their range is 0.15. (range is not the upper limit). If you want an upper limit of 0.15, then you should specify a range of 0.3

You will probably have to tweak your simulation quite a bit to make your 4-legged robot walk with the 4 stewart platforms. In any case you should try using the Vortex engine, which is the most robust among all 4 engines supported.

Cheers
In addition, I use this model http://pic.suiyiyun.cn/597454/6DOFStewart1-5.ttt to measure the forces of upper platform from 6 electrical cylinders. But druing the simulation, the joints have become ustable to connect. like this http://pic.suiyiyun.cn/597454/video.avi

Is my scene hierarchiy wrong?

Thanks a lot.

Post Reply