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

Guided Policy Search enables robots to learn control policies for complexmanipulation tasks efficiently. Therein, the control policies are representedas high-dimensional neural networks which derive robot actions based on states.However, due to the small number of real-world trajectory samples in GuidedPolicy Search, the resulting neural networks are only robust in theneighbourhood of the trajectory distribution explored by real-worldinteractions. In this paper, we present a new policy representation calledGenerative Motor Reflexes, which is able to generate robust actions over abroader state space compared to previous methods. In contrast to priorstate-action policies, Generative Motor Reflexes map states to parameters for astate-dependent motor reflex, which is then used to derive actions. Robustnessis achieved by generating similar motor reflexes for many states. We evaluatethe presented method in simulated and real-world manipulation tasks, includingcontact-rich peg-in-hole tasks. Using these evaluation tasks, we show thatpolicies represented as Generative Motor Reflexes lead to robust manipulationskills also outside the explored trajectory distribution with less trainingneeds compared to previous methods.

Learning Robust Manipulation Skills with Guided Policy Search via Generative Motor Reflexes