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

Recent works in robotic manipulation through reinforcement learning (RL) orimitation learning (IL) have shown potential for tackling a range of taskse.g., opening a drawer or a cupboard. However, these techniques generalizepoorly to unseen objects. We conjecture that this is due to thehigh-dimensional action space for joint control. In this paper, we take analternative approach and separate the task of learning 'what to do' from 'howto do it' i.e., whole-body control. We pose the RL problem as one ofdetermining the skill dynamics for a disembodied virtual manipulatorinteracting with articulated objects. The whole-body robotic kinematic controlis optimized to execute the high-dimensional joint motion to reach the goals inthe workspace. It does so by solving a quadratic programming (QP) model withrobotic singularity and kinematic constraints. Our experiments on manipulatingcomplex articulated objects show that the proposed approach is moregeneralizable to unseen objects with large intra-class variations,outperforming previous approaches. The evaluation results indicate that ourapproach generates more compliant robotic motion and outperforms the pure RLand IL baselines in task success rates. Additional information and videos areavailable at https://kl-research.github.io/decoupskill

Decoupling Skill Learning from Robotic Control for Generalizable Object Manipulation