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

Reinforcement learning methods can achieve significant performance butrequire a large amount of training data collected on the same robotic platform.A policy trained with expensive data is rendered useless after making even aminor change to the robot hardware. In this paper, we address the challengingproblem of adapting a policy, trained to perform a task, to a novel robotichardware platform given only few demonstrations of robot motion trajectories onthe target robot. We formulate it as a few-shot meta-learning problem where thegoal is to find a meta-model that captures the common structure shared acrossdifferent robotic platforms such that data-efficient adaptation can beperformed. We achieve such adaptation by introducing a learning frameworkconsisting of a probabilistic gradient-based meta-learning algorithm thatmodels the uncertainty arising from the few-shot setting with a low-dimensionallatent variable. We experimentally evaluate our framework on a simulatedreaching and a real-robot picking task using 400 simulated robots generated byvarying the physical parameters of an existing set of robotic platforms. Ourresults show that the proposed method can successfully adapt a trained policyto different robotic platforms with novel physical parameters and thesuperiority of our meta-learning algorithm compared to state-of-the-art methodsfor the introduced few-shot policy adaptation problem.

Bayesian Meta-Learning for Few-Shot Policy Adaptation Across Robotic Platforms