Training Classifiers For Feedback Control

@article{Poonawala2019TrainingCF,
  title={Training Classifiers For Feedback Control},
  author={Hasan A. Poonawala and Niklas T. Lauffer and Ufuk Topcu},
  journal={2019 American Control Conference (ACC)},
  year={2019},
  pages={4961-4967}
}
One approach for feedback control using high dimensional and rich sensor measurements is to classify the measurement into one out of a finite set of situations, each situation corresponding to a (known) control action. This approach computes a control action without estimating the state. Such classifiers are typically learned from a finite amount of data using supervised machine learning algorithms. We model the closed-loop system resulting from control with feedback from classifier outputs as… 

Figures from this paper

Training classifiers for feedback control with safety in mind
Stability Analysis of Conewise Affine Dynamical Systems Using Conewise Linear Lyapunov Functions
TLDR
It is shown that this approach verifies stability of 2D and 3D examples of conewise affine dynamical systems, including combinations of the harmonic and nonsmooth oscillators.
Stability Analysis Via Refinement Of Piece-wise Linear Lyapunov Functions
TLDR
This work presents an algorithm that involves iteratively refining partitions using an exact criterion which strictly reduces the set of points over which the Lyapunov function is non-decreasing.

References

SHOWING 1-10 OF 16 REFERENCES
Robustness of classifier-in-the-loop control systems: A hybrid-systems approach
TLDR
This paper studies continuous-state dynamical systems that use a classifier to determine the control input and proposes techniques to determine when the stability of the nominal system can be extended to unseen states and/or environments.
End-to-End Training of Deep Visuomotor Policies
TLDR
This paper develops a method that can be used to learn policies that map raw image observations directly to torques at the robot's motors, trained using a partially observed guided policy search method, with supervision provided by a simple trajectory-centric reinforcement learning method.
Nonquadratic Lyapunov functions for robust control
A Machine Learning Approach to Visual Perception of Forest Trails for Mobile Robots
TLDR
This work proposes a different approach to perceive forest trials based on a deep neural network used as a supervised image classifier that outperforms alternatives, and yields an accuracy comparable to the accuracy of humans that are tested on the same image classification task.
Switching Rule Design for Switched Dynamic Systems With Affine Vector Fields
TLDR
The method deals with the class of switched systems where each mode of operation is represented by a dynamical system with an affine vector field and it guarantees global asymptotic stability of the tracking error dynamics.
Introduction to machine learning
  • Ethem Alpaydin
  • Computer Science
    Adaptive computation and machine learning
  • 2004
TLDR
Introduction to Machine Learning is a comprehensive textbook on the subject, covering a broad array of topics not usually included in introductory machine learning texts, and discusses many methods from different fields, including statistics, pattern recognition, neural networks, artificial intelligence, signal processing, control, and data mining.
Support-Vector Networks
TLDR
High generalization ability of support-vector networks utilizing polynomial input transformations is demonstrated and the performance of the support- vector network is compared to various classical learning algorithms that all took part in a benchmark study of Optical Character Recognition.
Hybrid Dynamical Systems: Modeling, Stability, and Robustness
TLDR
This book presents a complete theory of robust asymptotic stability for hybrid dynamical systems that is applicable to the design of hybrid control algorithms--algorithms that feature logic, timers, or combinations of digital and analog components.
Piecewise Linear Control Systems
TLDR
This thesis treats analysis and design of piecewise linear control systems, and it is shown how Lyapunov functions with a discontinuous dependence on the discrete state can be computed via convex optimization.
Stabilization of Switched Systems via Composite Quadratic Functions
TLDR
This paper investigates stabilization of switched systems by using a Lyapunov function composed of a family of continuously differentiable functions, and it is observed that the min of quadratics, which is nondifferentiable and nonconvex, may be a more convenient tool than the other two types of functions which are convex and/or differentiable.
...
1
2
...