search

kristery / dex-affordance

[CoRL 2022] Implementation of "Learning Generalizable Dexterous Manipulation from Human Grasp Affordance"
https://kristery.github.io/ILAD/
38 stars 2 forks source link

Motion planning for grasp trajectory #2

Open LQTS opened 1 year ago

LQTS commented 1 year ago

Hi Yueh-Hua,

Thanks for providing the excellent project. It seems that I do not find the script of motion planning from a given grasping pose. Can you provide me with the script of CEM planner? Thank you very much!

Best, Quentin Liu

maple-zxh commented 4 months ago

Hi,did you solve this problem?

LQTS commented 4 months ago

Hi,did you solve this problem?

No, I have no idea. I tried many other methods, but I can't generate smooth trajectories. Maybe these methods can help you settle your problems down. If you want to try them, you can refer to https://pregrasps.github.io/ and https://github.com/aravindr93/trajopt

maple-zxh commented 4 months ago

Thank you for your reply, I will try them.

kristery commented 3 months ago

Hi Quentin Liu,

Thank you for your interest in our work. Here is our script for the cem planning: https://drive.google.com/drive/folders/1Cklymt7A9Ig3wPAw6BhNg8-UD022ZM9B?usp=drive_link

maple-zxh commented 3 months ago

Hi Yueh-Hua,

Thank you for the CEM planning script. If I want to replace it with my own model file, what I should provide is the pose (position, quaternion) of the object model as well as the poses corresponding to the 21 key points of the transformed hand, which are tested under the simulation environment. Is the initial pose of the object set randomly?

Best, Xuhui