A probabilistic data-driven model for planar pushing

@article{Bauz2017APD,
  title={A probabilistic data-driven model for planar pushing},
  author={Maria Bauz{\'a} and Alberto Rodriguez},
  journal={2017 IEEE International Conference on Robotics and Automation (ICRA)},
  year={2017},
  pages={3008-3015}
}
  • Maria Bauzá, Alberto Rodriguez
  • Published 10 April 2017
  • Computer Science, Mathematics, Engineering
  • 2017 IEEE International Conference on Robotics and Automation (ICRA)
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… Expand
Probabilistic modeling of planar pushing
TLDR
This approach can learn accurate data-driven models that outperform analytical models after less than 100 samples and saturate in performance with less than 1000 samples, and can predict heteroscedasticity, i.e., different amounts of uncertainty, and multi-modality when naturally occurring in pushing. Expand
Augmenting Physical Simulators with Stochastic Neural Networks: Case Study of Planar Pushing and Bouncing
TLDR
This paper builds an efficient, generalizable physical simulator with universal uncertainty estimates for two scenarios, planar pushing and ball bouncing, by augmenting an analytical rigid-body simulator with a neural network that learns to model uncertainty as residuals. Expand
Combining Analytical and Learned Models for Model Predictive Control
Robotics aims to deploy systems in increasingly complex and dynamic environments. This leads to a growing demand of robust and safe control algorithms for spatio-temporal domains, as failingExpand
Online Learning in Planar Pushing with Combined Prediction Model
TLDR
A combined prediction model and an online learning framework for planar push prediction that is able to quickly adapt to new environments, with the online prediction performance quickly converges to the level of offline performance. Expand
Self-Adapting Recurrent Models for Object Pushing from Learning in Simulation
TLDR
This paper proposes the algorithm Recurrent Model Predictive Path Integral (RMPPI) as a variation of the original MPPI approach, employing state-dependent recurrent models, and builds an LSTM-based model to fit the pushing dynamics. Expand
Friction Variability in Planar Pushing Data: Anisotropic Friction and Data-Collection Bias
TLDR
It is shown in this letter how a significant fraction of the observed unconventional phenomena, e.g., stochasticity and multimodality, can be explained by combining the effects of material nonhomogeneity, anisotropy of friction and biases due to data collection dynamics, suggesting that the variability is explainable but inevitable in practice. Expand
Long-Horizon Prediction and Uncertainty Propagation with Residual Point Contact Learners
TLDR
A self-supervised approach to learning residual models for rigid-body simulators that exploits corrections of contact models to refine predictive performance and propagate uncertainty is proposed. Expand
ON LEARNING HETEROSCEDASTIC NOISE MODELS
In many robotic applications, it is crucial to maintain a belief about the state of a system, like the location of a robot or the pose of an object. These state estimates serve as input for planningExpand
STReSSD: Sim-To-Real from Sound for Stochastic Dynamics
TLDR
STReSSD, a framework that uses sound to bridge the simulation-to-reality gap for stochastic dynamics, demonstrated for the canonical case of a bouncing ball, is presented. Expand
Feature-Based Transfer Learning for Robotic Push Manipulation
TLDR
This paper presents a data-efficient approach to learning transferable forward models for robotic push manipulation by leveraging information on the pushed object's local surface features, and test the hypothesis that this approach can achieve generalisation across objects of different shapes. Expand
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 25 REFERENCES
Most likely heteroscedastic Gaussian process regression
TLDR
This paper follows Goldberg et al.'s approach and model the noise variance using a second GP in addition to the GP governing the noise-free output value, using a Markov chain Monte Carlo method to approximate the posterior noise variance. Expand
More than a million ways to be pushed. A high-fidelity experimental dataset of planar pushing
TLDR
A comprehensive and high-fidelity dataset of planar pushing experiments is presented, characterizing the variability of friction, and evaluating the most common assumptions and simplifications made by models of frictional pushing in robotics. Expand
Gaussian Process Motion planning
TLDR
This paper develops the Gaussian Process Motion Planner (GPMP), a gradient-based optimization technique that optimizes continuous-time trajectories with respect to a cost functional by exploiting GP interpolation. Expand
Variational Heteroscedastic Gaussian Process Regression
TLDR
This work presents a non-standard variational approximation that allows accurate inference in heteroscedastic GPs (i.e., under input-dependent noise conditions) and its effectiveness is illustrated on several synthetic and real datasets of diverse characteristics. Expand
A Vision-Based Learning Method for Pushing Manipulation
We describe an unsupervised on-line method for learning of manipulative actions that allows a robot to push an object connected to it with a rotational point con- tact to a desired point inExpand
A convex polynomial force-motion model for planar sliding: Identification and application
TLDR
A simple and statistically-efficient model identification procedure using a sum-of-squares convex relaxation for planar sliding and practical applications including stable pushing of objects and free sliding dynamic simulations are shown. Expand
Push-manipulation of complex passive mobile objects using experimentally acquired motion models
TLDR
This work presents an experience-based push-manipulation approach that enables the robot to acquire experimental models regarding how pushable real world objects with complex 3D structures move in response to various pushing actions and demonstrates the superiority of the achievable planning and execution concept through safe and successful push- manipulation of a variety of passively mobile pushable objects. Expand
Sparse Gaussian Processes using Pseudo-inputs
TLDR
It is shown that this new Gaussian process (GP) regression model can match full GP performance with small M, i.e. very sparse solutions, and it significantly outperforms other approaches in this regime. Expand
Pushing using learned manipulation maps
  • Sean Walker, J. Salisbury
  • Engineering, Computer Science
  • 2008 IEEE International Conference on Robotics and Automation
  • 2008
TLDR
This paper describes one approach to modeling support friction within a two-dimensional environment consisting of a single robot finger pushing objects on a table using an online, memory-based model using local regression and describes the support friction without a complex model. Expand
Automatic learning of pushing strategy for delivery of irregular-shaped objects
TLDR
A learning-based approach for pushing objects of any irregular shape to user-specified goal locations by automatically collecting a set of data on how an irregular-shaped object moves given the robot's relative position and pushing direction. Expand
...
1
2
3
...