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

Self-supervised goal proposal and reaching is a key component for explorationand efficient policy learning algorithms. Such a self-supervised approachwithout access to any oracle goal sampling distribution requires deepexploration and commitment so that long horizon plans can be efficientlydiscovered. In this paper, we propose an exploration framework, which learns adynamics-aware manifold of reachable states. For a goal, our proposed methoddeterministically visits a state at the current frontier of reachable states(commitment/reaching) and then stochastically explores to reach the goal(exploration). This allocates exploration budget near the frontier of thereachable region instead of its interior. We target the challenging problem ofpolicy learning from initial and goal states specified as images, and do notassume any access to the underlying ground-truth states of the robot and theenvironment. To keep track of reachable latent states, we propose adistance-conditioned reachability network that is trained to infer whether onestate is reachable from another within the specified latent space distance.Given an initial state, we obtain a frontier of reachable states from thatstate. By incorporating a curriculum for sampling easier goals (closer to thestart state) before more difficult goals, we demonstrate that the proposedself-supervised exploration algorithm, superior performance compared toexisting baselines on a set of challenging roboticenvironments.https://sites.google.com/view/leaf-exploration

LEAF: Latent Exploration Along the Frontier