pytorch_kinematics icon indicating copy to clipboard operation
pytorch_kinematics copied to clipboard

jacobian error when build chain from mjcf

Open gautica opened this issue 4 months ago • 0 comments

Dear developer, Thank you very much for your great work. I have recently been using pytorch_kinematics to compute the Jacobian, and I found an issue that the Jacobian matrix is wrong when I build the chain from mjcf file, but it is correct when I build it from urdf. The problem is that the first 3 rows for linear velocity are always zeros. Here is an example: Jacobian when building a chain from MJCF: tensor([[[0., 0., 0., 0., 0., 0.], [0., 0., 0., 0., 0., 0.], [0., 0., 0., 0., 0., 0.], [0., 1., 0., 0., 1., 0.], [1., 0., 0., 1., 0., 1.], [0., 0., 1., 0., 0., 0.]]])

Jacoban using URDF: tensor([[[-7.4600e-01, 0.0000e+00, -4.8572e-16, -3.5700e-01, 0.0000e+00, 0.0000e+00], [ 0.0000e+00, 7.4600e-01, 2.0000e-02, 0.0000e+00, 2.2204e-16, 0.0000e+00], [-2.0000e-02, 4.5797e-16, 0.0000e+00, -4.0000e-02, 3.6082e-16, 0.0000e+00], [ 0.0000e+00, 1.0000e+00, 0.0000e+00, 0.0000e+00, 1.0000e+00, 0.0000e+00], [ 1.0000e+00, 0.0000e+00, 0.0000e+00, 1.0000e+00, 0.0000e+00, 1.0000e+00], [ 0.0000e+00, 0.0000e+00, 1.0000e+00, 0.0000e+00, 0.0000e+00, 0.0000e+00]]])

Thank you very much.

gautica avatar Aug 07 '25 14:08 gautica