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 ability to transfer a policy from one environment to another is apromising avenue for efficient robot learning in realistic settings where tasksupervision is not available. This can allow us to take advantage ofenvironments well suited for training, such as simulators or laboratories, tolearn a policy for a real robot in a home or office. To succeed, such policytransfer must overcome both the visual domain gap (e.g. different illuminationor background) and the dynamics domain gap (e.g. different robot calibration ormodelling error) between source and target environments. However, prior policytransfer approaches either cannot handle a large domain gap or can only addressone type of domain gap at a time. In this paper, we propose a novel policytransfer method with iterative "environment grounding", IDAPT, that alternatesbetween (1) directly minimizing both visual and dynamics domain gaps bygrounding the source environment in the target environment domains, and (2)training a policy on the grounded source environment. This iterative trainingprogressively aligns the domains between the two environments and adapts thepolicy to the target environment. Once trained, the policy can be directlyexecuted on the target environment. The empirical results on locomotion androbotic manipulation tasks demonstrate that our approach can effectivelytransfer a policy across visual and dynamics domain gaps with minimalsupervision and interaction with the target environment. Videos and code areavailable at https://clvrai.com/idapt .

Policy Transfer across Visual and Dynamics Domain Gaps via Iterative Grounding