The joint velocity (qdot) in stage 3 is always zero

[jgoenetxea]

Hi @soshishimada,

Thank you for sharing such an interesting piece of code with the research community. The captured motion looks amazing.

However, there is something that I do not understand in the implementation.

It is about the body joint velocity values computed in stage 3. In the code, the velocity for all the joints is always zero. However, the velocity looks like a key value to calculate the estimated pose in eq.(1) and the desired acceleration in eq.(4). It is also part of c matrix computation. Is this a negligible value?

Moreover, in the same function, the estimated body pose is computed with:

...
            """  Pose update """ 
            q =  pre_q + delta_t * qdot + acc*delta_t * delta_t/2
...

I assume that this is a combination of both functions included in eq. (1). However, the acceleration is divided by 2, but I do not where does it come from.

Kind regards.

[soshishimada]

Hi, sorry for the late reply. Somehow I didn't get a notification.

Apparently, I wrote the finite difference part wrongly when I was reimplementing PhysCap. It's updated now. Velocity is updated every iteration.