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/

Zendy Open

Presents the pdf analysis as premium feature

PDF Analysis

Insights

Presents the summarisation as premium feature

Summarisation

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the zaia usage as premium feature

ZAIA

Zendy Tools

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

This article presents a method for grasping novel objects by learning fromexperience. Successful attempts are remembered and then used to guide futuregrasps such that more reliable grasping is achieved over time. To generalisethe learned experience to unseen objects, we introduce the dense geometriccorrespondence matching network (DGCM-Net). This applies metric learning toencode objects with similar geometry nearby in feature space. Retrievingrelevant experience for an unseen object is thus a nearest neighbour searchwith the encoded feature maps. DGCM-Net also reconstructs 3D-3D correspondencesusing the view-dependent normalised object coordinate space to transform graspconfigurations from retrieved samples to unseen objects. In comparison tobaseline methods, our approach achieves an equivalent grasp success rate.However, the baselines are significantly improved when fusing the knowledgefrom experience with their grasp proposal strategy. Offline experiments with agrasping dataset highlight the capability to generalise within and betweenobject classes as well as to improve success rate over time from increasingexperience. Lastly, by learning task-relevant grasps, our approach canprioritise grasps that enable the functional use of objects.

DGCM-Net: Dense Geometrical Correspondence Matching Network for Incremental Experience-based Robotic Grasping