Deep reinforcement learning and randomized blending for control under novel disturbances
| dc.contributor.author | Sohége, Yves | |
| dc.contributor.author | Provan, Gregory | |
| dc.contributor.author | Quinones-Grueiro, Marcos | |
| dc.contributor.author | Biswas, Gautam | |
| dc.contributor.funder | Science Foundation Ireland | en |
| dc.date.accessioned | 2021-02-26T14:44:05Z | |
| dc.date.available | 2021-02-26T14:44:05Z | |
| dc.date.issued | 2020-07 | |
| dc.date.updated | 2021-02-25T15:26:16Z | |
| dc.description.abstract | Enabling autonomous vehicles to maneuver in novel scenarios is a key unsolved problem. A well-known approach, Weighted Multiple Model Adaptive Control (WMMAC), uses a set of pretuned controllers and combines their control actions using a weight vector. Although WMMAC offers an improvement to traditional switched control in terms of smooth control oscillations, it depends on accurate fault isolation and cannot deal with unknown disturbances. A recent approach avoids state estimation by randomly assigning the controller weighting vector; however, this approach uses a uniform distribution for control-weight sampling, which is sub-optimal compared to state-estimation methods. In this article, we propose a framework that uses deep reinforcement learning (DRL) to learn weighted control distributions that optimize the performance of the randomized approach for both known and unknown disturbances. We show that RL-based randomized blending dominates pure randomized blending, a switched FDI-based architecture and pre-tuned controllers on a quadcopter trajectory optimisation task in which we penalise deviations in both position and attitude. | en |
| dc.description.status | Peer reviewed | en |
| dc.description.version | Accepted Version | en |
| dc.format.mimetype | application/pdf | en |
| dc.identifier.citation | Sohege, Y., Provan, G., Quinones-Grueiro, M. and Biswas, G. (2020) 'Deep reinforcement learning and randomized blending for control under novel disturbances', IFAC World Congress 2020, Germany, 11-15 July. | en |
| dc.identifier.endpage | 6 | en |
| dc.identifier.startpage | 1 | en |
| dc.identifier.uri | https://hdl.handle.net/10468/11109 | |
| dc.language.iso | en | en |
| dc.relation.project | info:eu-repo/grantAgreement/SFI/SFI Research Centres/12/RC/2289/IE/INSIGHT - Irelands Big Data and Analytics Research Centre/ | en |
| dc.rights | © 2020 The authors | en |
| dc.rights.uri | https://creativecommons.org/licenses/by-nc-nd/3.0/ie/ | en |
| dc.subject | Design of fault tolerant/reliable systems | en |
| dc.subject | Fault accommodation and Reconfiguration strategies | en |
| dc.subject | Methods based on neural networks and/or fuzzy logic for FDI | en |
| dc.title | Deep reinforcement learning and randomized blending for control under novel disturbances | en |
| dc.type | Conference item | en |
