Non-linear trajectory issue

[wzx16]

Hi,

I'm using the attached mjcf to play with the examples/Mujoco/force_osc_xyz_linear_path_gaussian_velocity.py. But I found the trajectory of the end-effector is not linear as the figure shows. My questions are:

1. Is there anyway to fix this issue?
2. If it's impossible, I was wondering if you know any robotiq85 gripper working fine with the built-in ur5 of abr_control?
3. May I know the metrics used in this project?

eef traj:

mjcf xml: ur5e_gripper.zip

Thank you!

[p3jawors]

Hi,

When I run the sim the reach looks quite straight in its trajectory. It looks like it may be a plotting artefact. If you set your axes to be equal it should look better, if you look at your one axis the range is much smaller than the others.

If you add this: ax3.set_aspect('equal') before the plt.show() call you should get the following figure:

Let me know if this doesn't resolve the issue.

Cheers,

Pawel

[wzx16]

Hi Pawel, thank you for the answer. I don't think it's an artifact issue. Instead, I found if I call pp.next() for each 2 iterations in the loop, the problem resolved. However, the trajectory(blue) is a little bit off from the planned path(yellow) as the image shows. Could you provide any insight? And I'd like to double check if you are using the mjcf file I attached in last initial post? Thank you!

[quangminh2702]

hi @zxwang96. Could you please send me the config.py file for the UR5e robot? Thank you!

[wzx16]

Hi @quangminh2702, thanks for the reply. I've solved this issue. It turns out to be a mjcf issue, closing this issue.

[quangminh2702]

Hi @zxwang96. I am also trying to control the ur5e robot, but I do not have the config.py for that. So it would be great if you could send me the config file for the ur5e robot. Thank you!

[wzx16]

which config.py? I just use the mjcf file downloaded from mujoco_menagerie and modify the examples/Mujoco/force_osc_xyz_linear_path_gaussian_velocity.py file accordingly