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

Generating human-like behavior on robots is a great challenge especially indexterous manipulation tasks with robotic hands. Scripting policies fromscratch is intractable due to the high-dimensional control space, and trainingpolicies with reinforcement learning (RL) and manual reward engineering canalso be hard and lead to unnatural motions. Leveraging the recent progress onRL from Human Feedback, we propose a framework that learns a universal humanprior using direct human preference feedback over videos, for efficientlytuning the RL policies on 20 dual-hand robot manipulation tasks in simulation,without a single human demonstration. A task-agnostic reward model is trainedthrough iteratively generating diverse polices and collecting human preferenceover the trajectories; it is then applied for regularizing the behavior ofpolices in the fine-tuning stage. Our method empirically demonstrates morehuman-like behaviors on robot hands in diverse tasks including even unseentasks, indicating its generalization capability.

Learning a Universal Human Prior for Dexterous Manipulation from Human Preference