Physically Plausible Scene Estimation for Manipulation in Clutter

Desingh Karthik 1, 2 Chadwicke Odest 2 Lionel Reveret 3, 4 Sui Zhiqiang 1, 2
3 MORPHEO - Capture and Analysis of Shapes in Motion
Inria Grenoble - Rhône-Alpes, LJK - Laboratoire Jean Kuntzmann, INPG - Institut National Polytechnique de Grenoble
4 CVGI - Calcul des Variations, Géométrie, Image
LJK - Laboratoire Jean Kuntzmann
Abstract : Perceiving object poses in a cluttered scene is a challenging problem because of the partial observations available to an embodied robot, where cluttered scenes are especially problematic. In addition to occlusions, cluttered scenes have various cases of uncertainty due to physical object interactions, such as touching, stacking and partial support. In this paper, we discuss these cases of physics-based uncertainty case by case and propose methods for physically-viable scene estimation. Specifically, we use Newtonian physical simulation to check the plausibility of hypotheses within generative probabilistic inference in relation to particle filtering, MCMC and an MCMC variant on particle filtering. Assuming that object geometries are known, we estimate the scene as a collection of object poses, and infer a distribution over the state space as well as the maximu likelihood estimate. We compare with ICP based approaches and present our results for scene estimation in isolated cases of physical object interaction as well as multiobject scenes such that manipulation of graspable objects can be performed with a PR2 robot.
Document type :
Conference papers
Complete list of metadatas

https://hal.inria.fr/hal-01426371
Contributor : Lionel Reveret <>
Submitted on : Wednesday, January 4, 2017 - 2:22:08 PM
Last modification on : Wednesday, August 7, 2019 - 2:34:15 PM

Identifiers

Citation

Desingh Karthik, Chadwicke Odest, Lionel Reveret, Sui Zhiqiang. Physically Plausible Scene Estimation for Manipulation in Clutter. Humanoids 2016 - IEEE-RAS 16th International Conference on Humanoid Robots, Nov 2016, Cancun, Mexico. pp.1073-1080, ⟨10.1109/HUMANOIDS.2016.7803404⟩. ⟨hal-01426371⟩

Share

Metrics

Record views

262