dc.contributor.author | Løver, Jakob | |
dc.contributor.author | Gjærum, Vilde Benoni | |
dc.contributor.author | Lekkas, Anastasios M. | |
dc.date.accessioned | 2022-03-21T13:47:19Z | |
dc.date.available | 2022-03-21T13:47:19Z | |
dc.date.created | 2021-09-30T11:23:36Z | |
dc.date.issued | 2021 | |
dc.identifier.citation | IFAC-PapersOnLine. 2021, 54 (16), 146-152. | en_US |
dc.identifier.issn | 2405-8963 | |
dc.identifier.uri | https://hdl.handle.net/11250/2986554 | |
dc.description.abstract | Artifical neural networks (ANNs) have made their way into marine robotics in the last years, where they are used in control and perception systems, to name a few examples. At the same time, the black-box nature of ANNs is responsible for key challenges related to interpretability and trustworthiness, which need to be addressed if ANNs are to be deployed safely in real-life operations. In this paper, we implement three XAI methods to provide explanations to the decisions made by a deep reinforcement learning agent: Kernel SHAP, LIME and Linear Model Trees (LMTs). The agent was trained via Proximal Policy Optimization (PPO) to perform automatic docking on a fully-actuated vessel. We discuss the properties and suitability of the three methods, and juxtapose them with important attributes of the docking agent to provide context to the explanations. | en_US |
dc.language.iso | eng | en_US |
dc.publisher | Elsevier | en_US |
dc.rights | Attribution-NonCommercial-NoDerivatives 4.0 Internasjonal | * |
dc.rights.uri | http://creativecommons.org/licenses/by-nc-nd/4.0/deed.no | * |
dc.title | Explainable AI methods on a deep reinforcement learning agent for automatic docking | en_US |
dc.type | Peer reviewed | en_US |
dc.type | Journal article | en_US |
dc.description.version | publishedVersion | en_US |
dc.source.pagenumber | 146-152 | en_US |
dc.source.volume | 54 | en_US |
dc.source.journal | IFAC-PapersOnLine | en_US |
dc.source.issue | 16 | en_US |
dc.identifier.doi | 10.1016/j.ifacol.2021.10.086 | |
dc.identifier.cristin | 1941229 | |
dc.relation.project | Norges forskningsråd: 304843 | en_US |
dc.relation.project | Norges forskningsråd: 223254 | en_US |
cristin.ispublished | true | |
cristin.fulltext | original | |
cristin.qualitycode | 1 | |