DROP: Dexterous Reorientation via Online Planning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Li, Albert H., Culbertson, Preston, Kurtz, Vince, Ames, Aaron D.
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910858942087168
author Li, Albert H.
Culbertson, Preston
Kurtz, Vince
Ames, Aaron D.
author_facet Li, Albert H.
Culbertson, Preston
Kurtz, Vince
Ames, Aaron D.
contents Achieving human-like dexterity is a longstanding challenge in robotics, in part due to the complexity of planning and control for contact-rich systems. In reinforcement learning (RL), one popular approach has been to use massively-parallelized, domain-randomized simulations to learn a policy offline over a vast array of contact conditions, allowing robust sim-to-real transfer. Inspired by recent advances in real-time parallel simulation, this work considers instead the viability of online planning methods for contact-rich manipulation by studying the well-known in-hand cube reorientation task. We propose a simple architecture that employs a sampling-based predictive controller and vision-based pose estimator to search for contact-rich control actions online. We conduct thorough experiments to assess the real-world performance of our method, architectural design choices, and key factors for robustness, demonstrating that our simple sampling-based approach achieves performance comparable to prior RL-based works. Supplemental material: https://caltech-amber.github.io/drop.
format Preprint
id arxiv_https___arxiv_org_abs_2409_14562
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle DROP: Dexterous Reorientation via Online Planning
Li, Albert H.
Culbertson, Preston
Kurtz, Vince
Ames, Aaron D.
Robotics
Achieving human-like dexterity is a longstanding challenge in robotics, in part due to the complexity of planning and control for contact-rich systems. In reinforcement learning (RL), one popular approach has been to use massively-parallelized, domain-randomized simulations to learn a policy offline over a vast array of contact conditions, allowing robust sim-to-real transfer. Inspired by recent advances in real-time parallel simulation, this work considers instead the viability of online planning methods for contact-rich manipulation by studying the well-known in-hand cube reorientation task. We propose a simple architecture that employs a sampling-based predictive controller and vision-based pose estimator to search for contact-rich control actions online. We conduct thorough experiments to assess the real-world performance of our method, architectural design choices, and key factors for robustness, demonstrating that our simple sampling-based approach achieves performance comparable to prior RL-based works. Supplemental material: https://caltech-amber.github.io/drop.
title DROP: Dexterous Reorientation via Online Planning
topic Robotics
url https://arxiv.org/abs/2409.14562