A probabilistic data-driven model for planar pushing

By Maria Bauzá, Alberto Rodriguez. 2017

113 citations recorded by the source corpus.

This paper presents a data-driven approach to model planar pushing interaction to predict both the most likely outcome of a push and its expected variability. The learned models rely on a variation of Gaussian processes with input-dependent noise called Variational Heteroscedastic Gaussian processes (VHGP) [1] that capture the mean and variance of a stochastic function. We show that we can learn accurate models that outperform analytical models after less than 100 samples and saturate in performance with less than 1000 samples. We validate the results against a collected dataset of repeated trajectories, and use the learned models to study questions such as the nature of the variability in pushing, and the validity of the quasi-static assumption.

Subfields: Biomedical Engineering, Mechanics of Materials

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Biomedical Engineering

In Mechanics of Materials

Citations in the corpus

Learning Synergies Between Pushing and Grasping with Self-Supervised Deep Reinforcement Learning (2018, background): “for learning the dynamics of pushing [4], [18], many of these works have largely focused on the execution of stable pushes for one object at a time.”
ContactNets: Learning of Discontinuous Contact Dynamics with Smooth, Implicit Representations (2020, methodology): “Two ubiquitous representations, fully-connected deep neural networks (DNNs) and Gaussian processes regression with squaredexponential kernels [5, 6, 7], are biased towards similar interpretations of Occam’s razor: the best parameterization (i.”
Realtime State Estimation with Tactile and Visual Sensing. Application to Planar Manipulation (2017, background): “Bauza and Rodriguez [21] use a Gaussian process to learn a stochastic model directly from data without physical modeling.”
A Review of Robot Learning for Manipulation: Challenges, Representations, and Algorithms (2019, background): “For example, an accurate model for planar pushing of a block may be learned from less than a hundred samples [23].”
Learning Visual Affordances for Robotic Manipulation (2019, background): “These goals are difficult to define or reward for model-based [70, 35, 43] or data-driven [10, 28, 48] approaches.”