English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

For robotic decision-making under uncertainty, the balance betweenexploitation and exploration of available options must be carefully taken intoaccount. In this study, we introduce a new variant of contextual multi-armedbandits called observation-augmented CMABs (OA-CMABs) wherein a decision-makingagent can utilize extra outcome observations from an external informationsource. CMABs model the expected option outcomes as a function of contextfeatures and hidden parameters, which are inferred from previous optionoutcomes. In OA-CMABs, external observations are also a function of contextfeatures and thus provide additional evidence about the hidden parameters. Yet,if an external information source is error-prone, the resulting posteriorupdates can harm decision-making performance unless the presence of errors isconsidered. To this end, we propose a robust Bayesian inference process forOA-CMABs that is based on the concept of probabilistic data validation. Ourapproach handles complex mixture model parameter priors and hybrid observationlikelihoods for semantic data sources, allowing us to develop validationalgorithms based on recently develop probabilistic semantic data associationtechniques. Furthermore, to more effectively cope with the combined sources ofuncertainty in OA-CMABs, we derive a new active inference algorithm for optionselection based on expected free energy minimization. This generalizes previouswork on active inference for bandit-based robotic decision-making by accountingfor faulty observations and non-Gaussian inference. Our approaches aredemonstrated on a simulated asynchronous search site selection problem forspace exploration. The results show that even if incorrect observations areprovided by external information sources, efficient decision-making and robustparameter inference are still achieved in a wide variety of experimentalconditions.

Observation-Augmented Contextual Multi-Armed Bandits for Robotic Exploration with Uncertain Semantic Data