Simulate a Tube Robot

[LMI-pewpew]

Hi there,

Im searching for a way to simulate a concentric tube robot and I stumbled upon PyElastica.

Can I simulate tubes using this software? Its not clear to me that i can after going through your website.

This may be a very simple question, but I wanted to be sure before I dive deeper into PyElastica.

Thanks!

[skim0119]

@LMI-pewpew Hi,

Yes, you can simulate tubes. I'm also working on a softrobot project that uses a silicon tube, and I'm using PyElastica as a simulation tool. I use CosseratRod in the package and manually modify the area, bend_matrix, and shear_matrix.

In my experience, the simulation gives accurate kinematics as long as there is no buckling. Also, I wouldn't expect the change in radius to be accurate, since the original CosseratRod assumes a circular (not hollow) cross-section, but I found this is not significant for a mild deformation.

[LMI-pewpew]

Thanks @skim0119 for the very useful information,

What do you consider mild deformation? I am working with sub millimeter diameter tubes where the bend radii of the tubes range from 20 to 0 mm.

[skim0119]

I mean mild deformation as there is no buckling or any bifurcation (like collapsing). I don't have any quantitative study at this point, and it would vary from case to case.

I'm using a tube of a diameter 1 cm (20-30cm length) and bend radii of about 5 cm (I haven't tested further). The simulation result with PyElastica roughly match the experiment. I'm saying rougly because my instrument measuring the position has some error on its own (The instrument has 2-4% error, while the simulation matched within 5%; equivalent to maximum 0-0.5mm position error). Overall configuration (geometry and morphology) and kinematic matches. In my case, I activate the tube using the internal pressure (hydraulic and pneumatic), so it would have more resistance to the buckling behavior.

I would say the best approach is to build a quick simulation using built-in modules and see if the result approximately matches within the workspace of your robot. If the result is promising, refine the boundary conditions and forcing to match the experiment, and measure the precision.

[LMI-pewpew]

Thanks for all the info, much appreciated!!