problem in linking my model's pieces, physics simulation, re

[fferri]

hi all,
this is my first attempt in modeling a robot.
it is a tracked robot with 4 active flippers that can be rotated individually:

I try to put the pieces together, by copying the structure of components/locomotion/caterpillar model, thus using the trick of having some actuated cylinders respondable and not collidable, for achieving motion of the mobile base.
however the model exhibits some problems during the simulation, and moreover, I cannot get the laser and the body to stick with the rest of the construction.
what am I doing wrong?

this is my work in progress:
https://dl.dropboxusercontent.com/u/523 ... 130909.ttt

[coppelia]

Hello,

Make sure you have carefully read and well understood this page about the basic dynamics design considerations. Best would be to also entirelly try to follow the BubbleRob tutorial at least.

Cheers

[fferri]

thanks.
I added dynamic joints to link body to tracks (I needed them anyway, because the tracks are not fixed, but they rotate wrt the body), plus other fixes.



however, in the real robot, there is a differential lock, so the two track joints are constrained to have the same angle (same and opposite, if seen from the same side).

how to achieve such differential joint pair? If I put one of the joints in dependent mode, the model falls apart.

by the way, the model looks quite "elastic", resulting in an unrealistic simulation. how can I improve it?

[Eric]

Hi!

To achieve the differential joint pair, I think you ll have to make a script that reads one joint angle and force the other to the opposite angle. As suggested earlier, it is a very good idea to do the tutorials. You ll learn the basic on how to make the kind of script you ll need for your differential joint.

Cheers

Eric

[coppelia]

Hello,

I am afraid that the differential is not that easy to achieve: if you simply constraint one joint to follow the inverse of the other (as Eric suggested for instance), then only one joint moves freely. The other one only copies its position and operates as a motor. But it is feasable, have a look at this video-post.

Cheers

[Eric]

My bad... i thought it would be an active joint (i saw a video of Talos and believed it was an active joint)

[fferri]

how about using such a configuration?



that is: there is a dynamic joint linking the two bogies, and a dependent joint linking one bogie to the body, having angle = 0.5*dynamicBogiesJoint.angle ?

[coppelia]

That's also a possibility if you don't want a differential per se.

But your hierarchy doesn't look right. For instance LeftBogie is directly parented with Caterpillar_respondable (there is no joint in-between).

Try:

LeftBoggy
---> passiveJoint ---> RightBoggy
---> activeJoint ---> Caterpillar_respondable

Cheers