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

We present a framework for solving long-horizon planning problems involvingmanipulation of rigid objects that operates directly from a point-cloudobservation, i.e. without prior object models. Our method plans in the space ofobject subgoals and frees the planner from reasoning about robot-objectinteraction dynamics by relying on a set of generalizable manipulationprimitives. We show that for rigid bodies, this abstraction can be realizedusing low-level manipulation skills that maintain sticking contact with theobject and represent subgoals as 3D transformations. To enable generalizationto unseen objects and improve planning performance, we propose a novel way ofrepresenting subgoals for rigid-body manipulation and a graph-attention basedneural network architecture for processing point-cloud inputs. Weexperimentally validate these choices using simulated and real-worldexperiments on the YuMi robot. Results demonstrate that our method cansuccessfully manipulate new objects into target configurations requiringlong-term planning. Overall, our framework realizes the best of the worlds oftask-and-motion planning (TAMP) and learning-based approaches. Project website:https://anthonysimeonov.github.io/rpo-planning-framework/.

A Long Horizon Planning Framework for Manipulating Rigid Pointcloud Objects