We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:


References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Robotics

Title: Visual Identification of Articulated Object Parts

Abstract: As autonomous robots interact and navigate around real-world environments such as homes, it is useful to reliably identify and manipulate articulated objects, such as doors and cabinets. Many prior works in object articulation identification require manipulation of the object, either by the robot or a human. While recent works have addressed predicting articulation types from visual observations alone, they often assume prior knowledge of category-level kinematic motion models or sequence of observations where the articulated parts are moving according to their kinematic constraints. In this work, we propose FormNet, a neural network that identifies the articulation mechanisms between pairs of object parts from a single frame of an RGB-D image and segmentation masks. The network is trained on 100k synthetic images of 149 articulated objects from 6 categories. Synthetic images are rendered via a photorealistic simulator with domain randomization. Our proposed model predicts motion residual flows of object parts, and these flows are used to determine the articulation type and parameters. The network achieves an articulation type classification accuracy of 82.5% on novel object instances in trained categories. Experiments also show how this method enables generalization to novel categories and be applied to real-world images without fine-tuning.
Comments: Presented at the 2021 IEEE International Conference on Intelligent Robots and Systems (IROS 2021). This version contains the camera-ready paper with an appendix. Project page: this https URL
Subjects: Robotics (cs.RO)
Cite as: arXiv:2012.00284 [cs.RO]
  (or arXiv:2012.00284v3 [cs.RO] for this version)

Submission history

From: Jacky Liang [view email]
[v1] Tue, 1 Dec 2020 05:59:20 GMT (14534kb,D)
[v2] Sat, 10 Apr 2021 14:15:43 GMT (11921kb,D)
[v3] Thu, 9 Dec 2021 21:46:27 GMT (7050kb,D)

Link back to: arXiv, form interface, contact.