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

Object insertion is a classic contact-rich manipulation task. The taskremains challenging, especially when considering general objects of unknowngeometry, which significantly limits the ability to understand the contactconfiguration between the object and the environment. We study the problem ofaligning the object and environment with a tactile-based feedback insertionpolicy. The insertion process is modeled as an episodic policy that iteratesbetween insertion attempts followed by pose corrections. We explore differentmechanisms to learn such a policy based on Reinforcement Learning. The keycontribution of this paper is to demonstrate that it is possible to learn atactile insertion policy that generalizes across different object geometries,and an ablation study of the key design choices for the learning agent: 1) thetype of learning scheme: supervised vs. reinforcement learning; 2) the type oflearning schedule: unguided vs. curriculum learning; 3) the type of sensingmodality: force/torque (F/T) vs. tactile; and 4) the type of tactilerepresentation: tactile RGB vs. tactile flow. We show that the optimalconfiguration of the learning agent (RL + curriculum + tactile flow) exposed to4 training objects yields an insertion policy that inserts 4 novel objects withover 85.0% success rate and within 3~4 attempts. Comparisons between F/T andtactile sensing, shows that while an F/T-based policy learns more efficiently,a tactile-based policy provides better generalization.

Tactile-RL for Insertion: Generalization to Objects of Unknown Geometry