Compatibility with MX variables about adam HOT 8 CLOSED

Thank you I just tested both options and I was able to work with Mx variables

from adam.

CC @Giulero

from adam.

Hi @AFusco99! Can you provide a snippet of code on how you're using the library? Thanks!

from adam.

Yes, thanks

import casadi as cs
import numpy as np
model_path = "panda.urdf"
import adam
from adam.casadi import KinDynComputations

joints_name_list = [
           'panda_joint1', 'panda_joint2', 'panda_joint3', 'panda_joint4',
           'panda_joint5', 'panda_joint6', 'panda_joint7'
            ]
joint_angles=cs.MX.sym('q', 7)
root_link = 'panda_link0'
w_H_b = np.eye(4)
 kinDyn = KinDynComputations(model_path, joints_name_list, root_link)
frame="panda_hand"
Tmat=(kinDyn.forward_kinematics(frame,w_H_b,joint_angles))

This code is functional as long as joint_angles is defined as an SX symbolic variable however in my case I need to define it as an MX symbolic variable as I am using external functions and CasADI only supports MX variables in these cases.

from adam.

Got it! It shouldn't be possible to convert SX to MX and vice-versa (it seems open casadi/casadi#1870).

I created a branch with some modifications in order to let adam accept also MX types.

I opened this PR: #84. You might test changing the branch!

from adam.

Fyi: PR #84 merged

from adam.

Hi @AFusco99 ! I just noticed that you are using the forward_kinematics function.

I guess also that you could use the function forward_kinematics_fun function. This function returns the casadi function directly, and it should be possible to use it with SX, MX and DM (and numpy).

Your snippet would become

import casadi as cs
import numpy as np
model_path = "panda.urdf"
import adam
from adam.casadi import KinDynComputations

joints_name_list = [
           'panda_joint1', 'panda_joint2', 'panda_joint3', 'panda_joint4',
           'panda_joint5', 'panda_joint6', 'panda_joint7'
            ]
joint_angles=cs.MX.sym('q', 7)
root_link = 'panda_link0'
w_H_b = np.eye(4)
 kinDyn = KinDynComputations(model_path, joints_name_list, root_link)
frame="panda_hand"
T_mat_fun = kinDyn.forward_kinematics_fun(frame)
Tmat=T_mat_fun(w_H_b,joint_angles)

from adam.

@AFusco99, please have a look at #89. Anyway, using the *fun version of the function, you should be able to use MX!

from adam.