A System for General In-Hand Object Re-Orientation
By Tao Chen, Jie Xu, Pulkit Agrawal. Conference on Robot Learning, 2021
317 citations recorded by the source corpus.
In-hand object reorientation has been a challenging problem in robotics due to high dimensional actuation space and the frequent change in contact state between the fingers and the objects. We present a simple model-free framework that can learn to reorient objects with both the hand facing upwards and downwards. We demonstrate the capability of reorienting over 2000 geometrically different objects in both cases. The learned policies show strong zero-shot transfer performance on new objects. We provide evidence that these policies are amenable to real-world operation by distilling them to use observations easily available in the real world. The videos of the learned policies are available at: https://taochenshh.github.io/projects/in-hand-reorientation.
Subfields: Computer Vision and Pattern Recognition, Control and Systems Engineering, Human-Computer Interaction
View Paper: https://www.semanticscholar.org/paper/0ee9b633a0914b51f1eec3ad434752aa58e10149
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computer Vision and Pattern Recognition
- Jitendra Malik
- Luc Van Gool
- Trevor Darrell
- Dacheng Tao
- Xiaogang Wang
- Li Fei-Fei
- Ming-Hsuan Yang
- Cordelia Schmid
In Control and Systems Engineering
- Jitendra Malik
- Sergey Levine
- Dieter Fox
- Michael J. Black
- Trevor Darrell
- Animesh Garg
- Chelsea Finn
- Ilija Radosavovic
In Human-Computer Interaction
- Michael J. Black
- Dimitrios Tzionas
- Jitendra Malik
- Omid Taheri
- Ilija Radosavovic
- Trevor Darrell
- Dieter Fox
- Animesh Garg
Citations in the corpus
- Masked Visual Pre-training for Motor Control (2022, methodology): “We see that MoCo-v3 can achieve non-trivial performance, showing the generality of our approach.”
- GOAL: Generating 4D Whole-Body Motion for Hand-Object Grasping (2021, methodology): “To avoid big MoCap datasets, some methods use deep reinforcement learning (RL) for body-scene [5, 44, 45] or hand-object [6, 11] interactions.”
- Grasp'D: Differentiable Contact-rich Grasp Synthesis for Multi-fingered Hands (2022, background): “In addition, the captured grasps are only appropriate for human hands and not for robotic ones (which are important for many applications [1, 12]).”
- RVT: Robotic View Transformer for 3D Object Manipulation (2023, background): “A large volume of work achieves the multi-task generalization by using a generalizable task or action representation such as object point cloud [18, 19], semantic segmentation and optical flow [1], and object-centric representation [29, 30].”