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 advent of tactile sensors in robotics has sparked many ideas on howrobots can leverage direct contact measurements of their environmentinteractions to improve manipulation tasks. An important line of research inthis regard is that of grasp force control, which aims to manipulate objectssafely by limiting the amount of force exerted on the object. While prior workshave either hand-modeled their force controllers, employed model-basedapproaches, or have not shown sim-to-real transfer, we propose a model-freedeep reinforcement learning approach trained in simulation and then transferredto the robot without further fine-tuning. We therefore present a simulationenvironment that produces realistic normal forces, which we use to traincontinuous force control policies. An evaluation in which we compare against abaseline and perform an ablation study shows that our approach outperforms thehand-modeled baseline and that our proposed inductive bias and domainrandomization facilitate sim-to-real transfer. Code, models, and supplementaryvideos are available on https://sites.google.com/view/rl-force-ctrl

Towards Transferring Tactile-based Continuous Force Control Policies from Simulation to Robot