"Cubli" Simulation

[MeltingPlastic]

Could newton be used to simulate something like a "Cubli"



For example if there is a spinning wheel with sufficient mass inside the outer frame and a braking force or torque is applied - the outer frame would move as a result?

[Julio Jerez]

could you edit the post? I can't watch the video, I think the url is incorrect.

edit: never mind I edit it and now I can see it,
for what I can see this is a box with three spinning flying wheel on each size.

at first glance my guess Is that since the three fly wheels have symmetry the will generate gyro torque that will zero out the angular motion of the entire structure. That is my guess but gyro effect are center intrusive.

The answer is yes I believe Newton can simulated that, at least the new version that I am adding the ability to deal with gyro force as emerging form the kinematic derivative as opposed to before that the we added as external forces.

My only worry is that the can be a very no linear system, so integration error can grow up fast, so it may require a higher simulation rate, but let us see.

I will try to recreate it the weekend.

[Julio Jerez]

alter watching the video, I can tell you yes newton can do almos all the effects that the contraction show, specially the walk and jumps.

I see that the connotation is no passive, it have an active controller that Newton can also simulation using the Hinge Actuators,

all of the sting motion can be simulation with the current version, the self balancing is more tricky,
it is no clear if the controller has an active roll on eth self balancing because that device is not exactly a Gimbal which are device that do self balance in a passive way.

in any case if a cool challenge and I will try recreated it.

[MeltingPlastic]

That would be really cool to see,

I understand that there is a lot of control feedback going on that would need to be written separate from the pure simulation to actually control the movement of the box across a surface etc. But if you actually modeled the braking using friction with rigid bodies (or maybe just a counter torque?) and the outer frame moved as a result of the braking, that would be amazing to see.

I am starting to play with newton and I have just started to integrate it into the open source Urho3D: https://urho3d.github.io/ specifically my fork: https://github.com/TrevorCash/Urho3D.

Currently the engine uses Bullet which leaves much to be desired. But the engine has a really great outer structure and build system and makes it a good candidate I think for integration.

[Julio Jerez]

I understand that there is a lot of control feedback going on that would need to be written separate from the pure simulation to actually control the movement of the box across a surface etc. But if you actually modeled the braking using friction with rigid bodies (or maybe just a counter torque?) and the outer frame moved as a result of the braking, that would be amazing to see.

The effect should emerge from the simulation, or else would not be a simulation, it will be a clever animation.
The simulation should reproduce result that are very close to the real thing.
The Newton engine has all the functionality to recreate that device just like is described in the video and generate almost the same result.
Even the controller should be simulated more of less exactly the same way they do it.
For that we have joints like dCustomHingeActuator that is the equivalent of a robotic actuator, with this a step motor can be simulated to take the same inputs they are feeding to the step motors in the device, using these functions.

Code: Select all

dFloat GetActuatorAngle() const;
void SetTargetAngle(dFloat angle);
void SetAngularRate(dFloat rate);
void SetMaxTorque(dFloat torque);


I will make the a new demo tomorrow Cublic and see how far I get, then you can pick it from there.