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

The most data-efficient algorithms for reinforcement learning in robotics aremodel-based policy search algorithms, which alternate between learning adynamical model of the robot and optimizing a policy to maximize the expectedreturn given the model and its uncertainties. Among the few proposedapproaches, the recently introduced Black-DROPS algorithm exploits a black-boxoptimization algorithm to achieve both high data-efficiency and goodcomputation times when several cores are used; nevertheless, like allmodel-based policy search approaches, Black-DROPS does not scale to highdimensional state/action spaces. In this paper, we introduce a new modellearning procedure in Black-DROPS that leverages parameterized black-box priorsto (1) scale up to high-dimensional systems, and (2) be robust to largeinaccuracies of the prior information. We demonstrate the effectiveness of ourapproach with the "pendubot" swing-up task in simulation and with a physicalhexapod robot (48D state space, 18D action space) that has to walk forward asfast as possible. The results show that our new algorithm is moredata-efficient than previous model-based policy search algorithms (with andwithout priors) and that it can allow a physical 6-legged robot to learn newgaits in only 16 to 30 seconds of interaction time.

Using Parameterized Black-Box Priors to Scale Up Model-Based Policy Search for Robotics