dexsuite / dex-retargeting

https://yzqin.github.io/anyteleop/
MIT License
194 stars 23 forks source link

Curobo implementation #27

Open levi3001 opened 2 months ago

levi3001 commented 2 months ago

Thank you for your awesome work!!! From the paper, you state that:

Given the detected wrist and hand pose, our goal is to generate smooth and collision-free motion of robot arm to reach the target Cartesian end-effector pose. Real-time motion generation methods are required to have a smooth teleoperation experience. In the prior work of [16], the robot motion is driven by Riemannian Motion Policies (RMPs) [49, 7] that can calculate acceleration fields in real-time. However, accelerations towards a particular end- effector pose do not guarantee natural trajectories. In this work, we adopt CuRobo [56], a highly parallelized collision- free robot motion generation library accelerated by GPUs, to generate natural and reactive robot motion in real-time.

So, as I understand, there is a module to generate collision-free motion from the detected wrist and hand pose using the CuRobo framework. Have you integrated this module into your work? If you have, can you please kindly show me where it is?

Thank you.