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

Learning models of user behaviour is an important problem that is broadlyapplicable across many application domains requiring human-robot interaction.In this work, we show that it is possible to learn generative models fordistinct user behavioural types, extracted from human demonstrations, byenforcing clustering of preferred task solutions within the latent space. Weuse these models to differentiate between user types and to find cases withoverlapping solutions. Moreover, we can alter an initially guessed solution tosatisfy the preferences that constitute a particular user type bybackpropagating through the learned differentiable models. An advantage ofstructuring generative models in this way is that we can extract causalrelationships between symbols that might form part of the user's specificationof the task, as manifested in the demonstrations. We further parameterize thesespecifications through constraint optimization in order to find a safetyenvelope under which motion planning can be performed. We show that theproposed method is capable of correctly distinguishing between three usertypes, who differ in degrees of cautiousness in their motion, while performingthe task of moving objects with a kinesthetically driven robot in a tabletopenvironment. Our method successfully identifies the correct type, within thespecified time, in 99% [97.8 - 99.8] of the cases, which outperforms an IRLbaseline. We also show that our proposed method correctly changes a defaulttrajectory to one satisfying a particular user specification even with unseenobjects. The resulting trajectory is shown to be directly implementable on aPR2 humanoid robot completing the same task.

From Demonstrations to Task-Space Specifications: Using Causal Analysis to Extract Rule Parameterization from Demonstrations