Predictive Inverse Optimal Control for Linear-Quadratic-Gaussian Systems

Xiangli Chen, Brian Ziebart
Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics, PMLR 38:165-173, 2015.

Abstract

Predictive inverse optimal control is a powerful approach for estimating the control policy of an agent from observed control demonstrations. Its usefulness has been established in a number of large-scale sequential decision settings characterized by complete state observability. However, many real decisions are made in situations where the state is not fully known to the agent making decisions. Though extensions of predictive inverse optimal control to partially observable Markov decision processes have been developed, their applicability has been limited by the complexities of inference in those representations. In this work, we extend predictive inverse optimal control to the linear- quadratic-Gaussian control setting. We establish close connections between optimal control laws for this setting and the probabilistic predictions under our approach. We demonstrate the effectiveness and benefit in estimating control policies that are influenced by partial observability on both synthetic and real datasets.

Cite this Paper


BibTeX
@InProceedings{pmlr-v38-chen15d, title = {{Predictive Inverse Optimal Control for Linear-Quadratic-Gaussian Systems}}, author = {Xiangli Chen and Brian Ziebart}, booktitle = {Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics}, pages = {165--173}, year = {2015}, editor = {Guy Lebanon and S. V. N. Vishwanathan}, volume = {38}, series = {Proceedings of Machine Learning Research}, address = {San Diego, California, USA}, month = {09--12 May}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v38/chen15d.pdf}, url = { http://proceedings.mlr.press/v38/chen15d.html }, abstract = {Predictive inverse optimal control is a powerful approach for estimating the control policy of an agent from observed control demonstrations. Its usefulness has been established in a number of large-scale sequential decision settings characterized by complete state observability. However, many real decisions are made in situations where the state is not fully known to the agent making decisions. Though extensions of predictive inverse optimal control to partially observable Markov decision processes have been developed, their applicability has been limited by the complexities of inference in those representations. In this work, we extend predictive inverse optimal control to the linear- quadratic-Gaussian control setting. We establish close connections between optimal control laws for this setting and the probabilistic predictions under our approach. We demonstrate the effectiveness and benefit in estimating control policies that are influenced by partial observability on both synthetic and real datasets.} }
Endnote
%0 Conference Paper %T Predictive Inverse Optimal Control for Linear-Quadratic-Gaussian Systems %A Xiangli Chen %A Brian Ziebart %B Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2015 %E Guy Lebanon %E S. V. N. Vishwanathan %F pmlr-v38-chen15d %I PMLR %P 165--173 %U http://proceedings.mlr.press/v38/chen15d.html %V 38 %X Predictive inverse optimal control is a powerful approach for estimating the control policy of an agent from observed control demonstrations. Its usefulness has been established in a number of large-scale sequential decision settings characterized by complete state observability. However, many real decisions are made in situations where the state is not fully known to the agent making decisions. Though extensions of predictive inverse optimal control to partially observable Markov decision processes have been developed, their applicability has been limited by the complexities of inference in those representations. In this work, we extend predictive inverse optimal control to the linear- quadratic-Gaussian control setting. We establish close connections between optimal control laws for this setting and the probabilistic predictions under our approach. We demonstrate the effectiveness and benefit in estimating control policies that are influenced by partial observability on both synthetic and real datasets.
RIS
TY - CPAPER TI - Predictive Inverse Optimal Control for Linear-Quadratic-Gaussian Systems AU - Xiangli Chen AU - Brian Ziebart BT - Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics DA - 2015/02/21 ED - Guy Lebanon ED - S. V. N. Vishwanathan ID - pmlr-v38-chen15d PB - PMLR DP - Proceedings of Machine Learning Research VL - 38 SP - 165 EP - 173 L1 - http://proceedings.mlr.press/v38/chen15d.pdf UR - http://proceedings.mlr.press/v38/chen15d.html AB - Predictive inverse optimal control is a powerful approach for estimating the control policy of an agent from observed control demonstrations. Its usefulness has been established in a number of large-scale sequential decision settings characterized by complete state observability. However, many real decisions are made in situations where the state is not fully known to the agent making decisions. Though extensions of predictive inverse optimal control to partially observable Markov decision processes have been developed, their applicability has been limited by the complexities of inference in those representations. In this work, we extend predictive inverse optimal control to the linear- quadratic-Gaussian control setting. We establish close connections between optimal control laws for this setting and the probabilistic predictions under our approach. We demonstrate the effectiveness and benefit in estimating control policies that are influenced by partial observability on both synthetic and real datasets. ER -
APA
Chen, X. & Ziebart, B.. (2015). Predictive Inverse Optimal Control for Linear-Quadratic-Gaussian Systems. Proceedings of the Eighteenth International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 38:165-173 Available from http://proceedings.mlr.press/v38/chen15d.html .

Related Material