Issue with running soft robotics tutorial

[Leon-0919]

Hi @hugtalbot @bakpaul ,

Thanks for your previous help.

I am currently running a softrobotics tutorial and have a question with the settings. Much appreciate your explanation.

The torque can be calculated using the spring stiffness and the angle difference between rest and current state, but this can not be modified/tuned for specific application.

So is there a parameter in the that determines the torque output?

Please let me know if there is any confusion.

Best

[bakpaul]

Hello,

This is not an issue. The RestShapeSpringsForceField acts as a weak link between the actuator and the body. You can imagine it as a soft actuator with a spring/based link. The motor imposes an angle of pi/6 but then the attached body is linked to it through an angular spring, leaving some level of compliance to the joint.

This is one way of imposing this rotation, we call this a weak constraint. A stiffer way of constraining it would be to use a Lagrangian based constraint. This can be done using a JointConstraint. This will compute the real torque required to impose this pi/6 angle displacement, taking into account the contact and model deformation (but not the motor dynamic). An example is provided here : https://github.com/SofaDefrost/SoftRobots/blob/master/examples/component/constraint/JointConstraint/JointConstraint.py

[Leon-0919]

@bakpaul Thanks a lot for your quick response!

I get your point now. By the way, for the RestShapeSpringsForceField, do you modify the parameter stiffness or angular stiffness to vary the property of virtual angular spring? I can see stiffness is widely used in the tutorial online.

Additionally, I am trying the stiffer constraint as you suggest, I have the error
Python exception: ValueError: Object type JointConstraint<Vec1d> was not created The object 'JointConstraint' is not in the factory. But the following exits: : SlidingConstraint (75% match) : FixedConstraint (73% match) : ProjectToPointConstraint (71% match) : AttachConstraint (70% match) : ParabolicConstraint (70% match) : StopperConstraint (68% match)

I assume its because I am using sofa22.04 and I haven't updated my local softrobotics plugin since early 2023.
Just to clarify, do I need to get the latest softrobotics plugin to implement the jointconstraint?

Thanks again for your help.

Best

[bakpaul]

Yes, it was released early 2024. So you'll certainly also need to update your SOFA version

[Leon-0919]

hey @bakpaul , I am trying to implement the 24.06 version and I have some errors which I believe comes from the sofa version upgration.

for example, [ERROR] [PluginManager] Plugin not found: "SofaMeshCollision". I had this components in version 22.04 for collision detection and response, but I could not find it in the 24.06 component folder.

Additionally, I am still trying to find alternatives in version 22.04, I am wondering if there any other way of implementing the input. For example, how we could give the soft actuator an input force or torque from the base? Do you know which forcefield should I use? https://github.com/sofa-framework/sofa/tree/master/examples/Component I am not sure about whether the forcefield from sofa or softrobotics plugin would work better for this problem.

Any suggestion would be much appreciated.
Best

[hugtalbot]

Hi @Leon-0919

An alternative before moving to a JointConstraint as suggested by @bakpaul was to simply increase the stiffness of your weak constraint imposed by the RestShapeSpringsForceField. It should bring you closer to the optimal angle.

Else, Paul was right using a proper Constraint might work better.

Regarding your last post @Leon-0919, the following error [ERROR] [PluginManager] Plugin not found: "SofaMeshCollision" is just due to the fact that the library SofaMeshCollision was split and refactored in the NG refactoring project. So no worries, check however if all components got correctly created

Cheers

[Leon-0919]

@hugtalbot @bakpaul thanks for getting back to me!

I am having difficulties running binary 24.06 due to gcc version, so I would update jointconstraint result once I get it working.

As per your suggestion of increasing the stiffness value, I have a doubt about this. It can help achieve the desired angle (pi/6), however how would you calculate the torque implemented by the motor then?

From what I understand and implement, the output force/torque was calculated using the product of stiffness with angle differences (rest angle & free motion angle). When the stiffness is big enough, angle differences become 0..

Is there anything I am missing regarding the torque calculation? BTW, what does this force mean in the above animation?

Thanks again for your help.

Best