search

tud-airlab / mppi-isaac

Model Predictive Path Integral Control using isaacgym for rollouts, gpu-accelerated
166 stars 19 forks source link

Unable to achieve good grasping and pushing effects #55

Open zhe-beginner opened 2 months ago

zhe-beginner commented 2 months ago

Hello, thank you very much for considering my difficulties, I ran the "omnipanda_isaacgym_client.py" and "omnipanda_isaacgym_server.py". The robot can get close to the table, but the gripper can't grab the object, the gripper is shaking near the object and not making contact, what is the reason for this? Do we need to tweak and find a more suitable cost function? The same issue also occurs in "panda_push_client.py" and "panda_push_server.py". In most cases, robots cannot correctly push blocks, and sometimes they cannot even touch objects. Is there any way to improve the success rate. I really want to replicate your open-source technology. If I could get your help, I would be very grateful.

cpezzato commented 2 months ago

Hi @zhe-beginner , I checked and indeed we have some bugs to solve. While we work on it, if you want to reproduce the exact behavior as in the gif in the readme, you can git checkout v.0.1.0. That should work as expected.