DISCO: Double Likelihood-free Inference Stochastic Control

@article{Barcelos2020DISCODL,
  title={DISCO: Double Likelihood-free Inference Stochastic Control},
  author={Lucas Gomes Barcelos and Rafael Oliveira and Rafael Possas and Lionel Ott and Fabio Ramos},
  journal={2020 IEEE International Conference on Robotics and Automation (ICRA)},
  year={2020},
  pages={10969-10975}
}
Accurate simulation of complex physical systems enables the development, testing, and certification of control strategies before they are deployed into the real systems. As simulators become more advanced, the analytical tractability of the differential equations and associated numerical solvers incorporated in the simulations diminishes, making them difficult to analyse. A potential solution is the use of probabilistic inference to assess the uncertainty of the simulation parameters given real… 
View on IEEE
arxiv.org
8 Citations
Background Citations
4
Methods Citations
3

Figures and Tables from this paper

8 Citations

Dual Online Stein Variational Inference for Control and Dynamics

An implicit variational inference algorithm able to estimate distributions over model parameters and control inputs on-the-fly on the fly is devised, enabling the approximation of complex multi-modal posterior distributions, typically occurring in challenging and realistic robot navigation tasks.

Online BayesSim for Combined Simulator Parameter Inference and Policy Improvement

Experimental results suggest that both model-free reinforcement learning and model-based control strategies have better performance when compared to traditional domain randomization methods.

DiffCloud: Real-to-Sim from Point Clouds with Differentiable Simulation and Rendering of Deformable Objects

Experiments with highly deformable objects show that the proposed method can achieve comparable or better alignment with real object behavior, while reducing the time needed to achieve this by more than an order of magnitude.

A Bayesian Treatment of Real-to-Sim for Deformable Object Manipulation

This letter forms the real-to-sim problem as probabilistic inference over simulation parameters of deformable objects as well as a principled way to incorporate noisy state observations into modern Bayesian tools for simulation parameter inference.

Robot Learning From Randomized Simulations: A Review

A comprehensive review of sim-to-real research for robotics, focusing on a technique named “domain randomization” which is a method for learning from randomized simulations.

A Differentiable Newton-Euler Algorithm for Real-World Robotics

A differentiable simulator that can be used to identify the system parameters of real-world mechanical systems with complex friction models, holonomic as well as non-holonomic constraints and can be applied to a class of dynamical systems and guarantees physically plausible predictions.

BayesSimIG: Scalable Parameter Inference for Adaptive Domain Randomization with IsaacGym

A library that provides an implementation of BayesSim integrated with the recently released NVIDIA IsaacGym that allows large-scale parameter inference with end-toend GPU acceleration and provides an integration with TensorBoard to easily visualize slices of high-dimensional posteriors.

Neural Posterior Domain Randomization

NPDR is introduced, an algorithm that alternates between learning a policy from a randomized simu lator and adapting the posterior distribution over the simulator’s parameters in a Bayesian fashion and can learn transferable policies using fewer real-world rollouts than comparable algorithms.

References

SHOWING 1-10 OF 26 REFERENCES

BayesSim: adaptive domain randomization via probabilistic inference for robotics simulators

Results show that the posterior computed from BayesSim can be used for domain randomization outperforming alternative methods that randomize based on uniform priors.

Robust nonlinear model predictive control with reduction of uncertainty via dual control

Dual control is a technique that solves the tradeoff between using the input signal for the excitation of the system excitation signal (probing actions) and controlling it, which results in a better…

Sigma-point kalman filters for probabilistic inference in dynamic state-space models

This work has consistently shown that there are large performance benefits to be gained by applying Sigma-Point Kalman filters to areas where EKFs have been used as the de facto standard in the past, as well as in new areas where the use of the EKF is impossible.

Stochastic Nonlinear Model Predictive Control with State Estimation by Incorporation of the Unscented Kalman Filter

Nonlinear model predictive control has become a popular approach to deal with highly nonlinear and unsteady state systems, the performance of which can however deteriorate due to unaccounted…

Stochastic Model Predictive Control: An Overview and Perspectives for Future Research

Model predictive control (MPC) has demonstrated exceptional success for the high-performance control of complex systems. The conceptual simplicity of MPC as well as its ability to effectively cope…

Uncertainty averse pushing with model predictive path integral control

This work shows how to achieve this using a forward model learned from robot data to plan push manipulations using Gaussian Process Regression and an Ensemble of Mixture Density Networks that give estimates of the uncertainty in their predictions.

An integrated system for real-time model predictive control of humanoid robots

An integrated system based on real-time model-predictive control (MPC) applied to the full dynamics of the robot, which is possible due to the speed of the new physics engine (MuJoCo), the efficiency of the trajectory optimization algorithm, and the contact smoothing methods developed for the purpose of control optimization.

Learning Without Mixing: Towards A Sharp Analysis of Linear System Identification

It is proved that the ordinary least-squares (OLS) estimator attains nearly minimax optimal performance for the identification of linear dynamical systems from a single observed trajectory, and generalizes the technique to provide bounds for a more general class of linear response time-series.

Parameterized Tube Model Predictive Control

This paper develops a parameterized tube model predictive control (MPC) synthesis method that optimizes parameterized, Minkowski decomposable, state and control tubes and an associated, fully separable, nonlinear, control policy.

Information-Theoretic Model Predictive Control: Theory and Applications to Autonomous Driving

An information-theoretic approach to stochastic optimal control problems that can be used to derive general sampling-based optimization schemes is presented and applied to the task of aggressive autonomous driving around a dirt test track.