Object shape estimation and modeling, based on sparse Gaussian process implicit surfaces, combining visual data and tactile exploration

Research output: Contribution to journalArticlepeer-review

Standard

Object shape estimation and modeling, based on sparse Gaussian process implicit surfaces, combining visual data and tactile exploration. / Gandler, Gabriela Zarzar; Ek, Carl Henrik; Björkman, Mårten; Stolkin, Rustam; Bekiroglu, Yasemin.

In: Robotics and Autonomous Systems, Vol. 126, 103433, 01.04.2020, p. 1-16.

Research output: Contribution to journalArticlepeer-review

Harvard

APA

Vancouver

Author

Bibtex

@article{958eaca26f944520bc2b429a749395a9,
title = "Object shape estimation and modeling, based on sparse Gaussian process implicit surfaces, combining visual data and tactile exploration",
abstract = "Inferring and representing three-dimensional shapes is an important part of robotic perception. However, it is challenging to build accurate models of novel objects based on real sensory data, because observed data is typically incomplete and noisy. Furthermore, imperfect sensory data suggests that uncertainty about shapes should be explicitly modeled during shape estimation. Such uncertainty models can usefully enable exploratory action planning for maximum information gain and efficient use of data. This paper presents a probabilistic approach for acquiring object models, based on visual and tactile data. We study Gaussian Process Implicit Surface (GPIS) representation. GPIS enables a non-parametric probabilistic reconstruction of object surfaces from 3D data points, while also providing a principled approach to encode the uncertainty associated with each region of the reconstruction. We investigate different configurations for GPIS, and interpret an object surface as the level-set of an underlying sparse GP. Experiments are performed on both synthetic data, and also real data sets obtained from two different robots physically interacting with objects. We evaluate performance by assessing how close the reconstructed surfaces are to ground-truth object models. We also evaluate how well objects from different categories are clustered, based on the reconstructed surface shapes. Results show that sparse GPs enable a reliable approximation to the full GP solution, and the proposed method yields adequate surface representations to distinguish objects. Additionally the presented approach is shown to provide computational efficiency, and also efficient use of the robot{\textquoteright}s exploratory actions.",
keywords = "tactile sensing, shape modeling, implicit surface, 3D reconstruction, Gaussian process, regression",
author = "Gandler, {Gabriela Zarzar} and Ek, {Carl Henrik} and M{\aa}rten Bj{\"o}rkman and Rustam Stolkin and Yasemin Bekiroglu",
year = "2020",
month = apr,
day = "1",
doi = "10.1016/j.robot.2020.103433",
language = "English",
volume = "126",
pages = "1--16",
journal = "Robotics and Autonomous Systems",
issn = "0921-8890",
publisher = "Elsevier",

}

RIS

TY - JOUR

T1 - Object shape estimation and modeling, based on sparse Gaussian process implicit surfaces, combining visual data and tactile exploration

AU - Gandler, Gabriela Zarzar

AU - Ek, Carl Henrik

AU - Björkman, Mårten

AU - Stolkin, Rustam

AU - Bekiroglu, Yasemin

PY - 2020/4/1

Y1 - 2020/4/1

N2 - Inferring and representing three-dimensional shapes is an important part of robotic perception. However, it is challenging to build accurate models of novel objects based on real sensory data, because observed data is typically incomplete and noisy. Furthermore, imperfect sensory data suggests that uncertainty about shapes should be explicitly modeled during shape estimation. Such uncertainty models can usefully enable exploratory action planning for maximum information gain and efficient use of data. This paper presents a probabilistic approach for acquiring object models, based on visual and tactile data. We study Gaussian Process Implicit Surface (GPIS) representation. GPIS enables a non-parametric probabilistic reconstruction of object surfaces from 3D data points, while also providing a principled approach to encode the uncertainty associated with each region of the reconstruction. We investigate different configurations for GPIS, and interpret an object surface as the level-set of an underlying sparse GP. Experiments are performed on both synthetic data, and also real data sets obtained from two different robots physically interacting with objects. We evaluate performance by assessing how close the reconstructed surfaces are to ground-truth object models. We also evaluate how well objects from different categories are clustered, based on the reconstructed surface shapes. Results show that sparse GPs enable a reliable approximation to the full GP solution, and the proposed method yields adequate surface representations to distinguish objects. Additionally the presented approach is shown to provide computational efficiency, and also efficient use of the robot’s exploratory actions.

AB - Inferring and representing three-dimensional shapes is an important part of robotic perception. However, it is challenging to build accurate models of novel objects based on real sensory data, because observed data is typically incomplete and noisy. Furthermore, imperfect sensory data suggests that uncertainty about shapes should be explicitly modeled during shape estimation. Such uncertainty models can usefully enable exploratory action planning for maximum information gain and efficient use of data. This paper presents a probabilistic approach for acquiring object models, based on visual and tactile data. We study Gaussian Process Implicit Surface (GPIS) representation. GPIS enables a non-parametric probabilistic reconstruction of object surfaces from 3D data points, while also providing a principled approach to encode the uncertainty associated with each region of the reconstruction. We investigate different configurations for GPIS, and interpret an object surface as the level-set of an underlying sparse GP. Experiments are performed on both synthetic data, and also real data sets obtained from two different robots physically interacting with objects. We evaluate performance by assessing how close the reconstructed surfaces are to ground-truth object models. We also evaluate how well objects from different categories are clustered, based on the reconstructed surface shapes. Results show that sparse GPs enable a reliable approximation to the full GP solution, and the proposed method yields adequate surface representations to distinguish objects. Additionally the presented approach is shown to provide computational efficiency, and also efficient use of the robot’s exploratory actions.

KW - tactile sensing

KW - shape modeling

KW - implicit surface

KW - 3D reconstruction

KW - Gaussian process

KW - regression

U2 - 10.1016/j.robot.2020.103433

DO - 10.1016/j.robot.2020.103433

M3 - Article

VL - 126

SP - 1

EP - 16

JO - Robotics and Autonomous Systems

JF - Robotics and Autonomous Systems

SN - 0921-8890

M1 - 103433

ER -