Bayesian Nonparametric Adaptive Control Using Gaussian Processes
Most current model reference adaptive control (MRAC) methods rely on parametric adaptive elements, in which the number of parameters of the adaptive element are fixed a priori, often through expert judgment. An example of such an adaptive element is radial basis function networks (RBFNs), with RBF c...
| Main Authors: | , , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | en_US |
| Published: |
Institute of Electrical and Electronics Engineers (IEEE)
2015
|
| Online Access: | http://hdl.handle.net/1721.1/97050 https://orcid.org/0000-0001-8576-1930 |
| _version_ | 1826202489115901952 |
|---|---|
| author | Chowdhary, Girish Kingravi, Hassan A. How, Jonathan P. Vela, Patricio A. |
| author2 | Massachusetts Institute of Technology. Department of Aeronautics and Astronautics |
| author_facet | Massachusetts Institute of Technology. Department of Aeronautics and Astronautics Chowdhary, Girish Kingravi, Hassan A. How, Jonathan P. Vela, Patricio A. |
| author_sort | Chowdhary, Girish |
| collection | MIT |
| description | Most current model reference adaptive control (MRAC) methods rely on parametric adaptive elements, in which the number of parameters of the adaptive element are fixed a priori, often through expert judgment. An example of such an adaptive element is radial basis function networks (RBFNs), with RBF centers preallocated based on the expected operating domain. If the system operates outside of the expected operating domain, this adaptive element can become noneffective in capturing and canceling the uncertainty, thus rendering the adaptive controller only semiglobal in nature. This paper investigates a Gaussian process-based Bayesian MRAC architecture (GP-MRAC), which leverages the power and flexibility of GP Bayesian nonparametric models of uncertainty. The GP-MRAC does not require the centers to be preallocated, can inherently handle measurement noise, and enables MRAC to handle a broader set of uncertainties, including those that are defined as distributions over functions. We use stochastic stability arguments to show that GP-MRAC guarantees good closed-loop performance with no prior domain knowledge of the uncertainty. Online implementable GP inference methods are compared in numerical simulations against RBFN-MRAC with preallocated centers and are shown to provide better tracking and improved long-term learning. |
| first_indexed | 2024-09-23T12:07:57Z |
| format | Article |
| id | mit-1721.1/97050 |
| institution | Massachusetts Institute of Technology |
| language | en_US |
| last_indexed | 2024-09-23T12:07:57Z |
| publishDate | 2015 |
| publisher | Institute of Electrical and Electronics Engineers (IEEE) |
| record_format | dspace |
| spelling | mit-1721.1/970502022-09-28T00:21:28Z Bayesian Nonparametric Adaptive Control Using Gaussian Processes Chowdhary, Girish Kingravi, Hassan A. How, Jonathan P. Vela, Patricio A. Massachusetts Institute of Technology. Department of Aeronautics and Astronautics Massachusetts Institute of Technology. Laboratory for Information and Decision Systems How, Jonathan P. Most current model reference adaptive control (MRAC) methods rely on parametric adaptive elements, in which the number of parameters of the adaptive element are fixed a priori, often through expert judgment. An example of such an adaptive element is radial basis function networks (RBFNs), with RBF centers preallocated based on the expected operating domain. If the system operates outside of the expected operating domain, this adaptive element can become noneffective in capturing and canceling the uncertainty, thus rendering the adaptive controller only semiglobal in nature. This paper investigates a Gaussian process-based Bayesian MRAC architecture (GP-MRAC), which leverages the power and flexibility of GP Bayesian nonparametric models of uncertainty. The GP-MRAC does not require the centers to be preallocated, can inherently handle measurement noise, and enables MRAC to handle a broader set of uncertainties, including those that are defined as distributions over functions. We use stochastic stability arguments to show that GP-MRAC guarantees good closed-loop performance with no prior domain knowledge of the uncertainty. Online implementable GP inference methods are compared in numerical simulations against RBFN-MRAC with preallocated centers and are shown to provide better tracking and improved long-term learning. United States. Office of Naval Research. Multidisciplinary University Research Initiative (Grant N000141110688) National Science Foundation (U.S.) (Grant ECS 0846750) 2015-05-21T14:44:34Z 2015-05-21T14:44:34Z 2015-02 2013-02 Article http://purl.org/eprint/type/JournalArticle 2162-237X 2162-2388 http://hdl.handle.net/1721.1/97050 Chowdhary, Girish, Hassan A. Kingravi, Jonathan P. How, and Patricio A. Vela. “Bayesian Nonparametric Adaptive Control Using Gaussian Processes.” IEEE Transactions on Neural Networks and Learning Systems 26, no. 3 (March 2015): 537–550. https://orcid.org/0000-0001-8576-1930 en_US http://dx.doi.org/10.1109/TNNLS.2014.2319052 IEEE Transactions on Neural Networks and Learning Systems Creative Commons Attribution-Noncommercial-Share Alike http://creativecommons.org/licenses/by-nc-sa/4.0/ application/pdf Institute of Electrical and Electronics Engineers (IEEE) MIT web domain |
| spellingShingle | Chowdhary, Girish Kingravi, Hassan A. How, Jonathan P. Vela, Patricio A. Bayesian Nonparametric Adaptive Control Using Gaussian Processes |
| title | Bayesian Nonparametric Adaptive Control Using Gaussian Processes |
| title_full | Bayesian Nonparametric Adaptive Control Using Gaussian Processes |
| title_fullStr | Bayesian Nonparametric Adaptive Control Using Gaussian Processes |
| title_full_unstemmed | Bayesian Nonparametric Adaptive Control Using Gaussian Processes |
| title_short | Bayesian Nonparametric Adaptive Control Using Gaussian Processes |
| title_sort | bayesian nonparametric adaptive control using gaussian processes |
| url | http://hdl.handle.net/1721.1/97050 https://orcid.org/0000-0001-8576-1930 |
| work_keys_str_mv | AT chowdharygirish bayesiannonparametricadaptivecontrolusinggaussianprocesses AT kingravihassana bayesiannonparametricadaptivecontrolusinggaussianprocesses AT howjonathanp bayesiannonparametricadaptivecontrolusinggaussianprocesses AT velapatricioa bayesiannonparametricadaptivecontrolusinggaussianprocesses |