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

Research in manipulation of deformable objects is typically conducted on alimited range of scenarios, because handling each scenario on hardware takessignificant effort. Realistic simulators with support for various types ofdeformations and interactions have the potential to speed up experimentationwith novel tasks and algorithms. However, for highly deformable objects it ischallenging to align the output of a simulator with the behavior of realobjects. Manual tuning is not intuitive, hence automated methods are needed. Weview this alignment problem as a joint perception-inference challenge anddemonstrate how to use recent neural network architectures to successfullyperform simulation parameter inference from real point clouds. We analyze theperformance of various architectures, comparing their data and trainingrequirements. Furthermore, we propose to leverage differentiable point cloudsampling and differentiable simulation to significantly reduce the time toachieve the alignment. We employ an efficient way to propagate gradients frompoint clouds to simulated meshes and further through to the physical simulationparameters, such as mass and stiffness. Experiments with highly deformableobjects show that our method can achieve comparable or better alignment withreal object behavior, while reducing the time needed to achieve this by morethan an order of magnitude. Videos and supplementary material are available athttps://tinyurl.com/diffcloud.

DiffCloud: Real-to-Sim from Point Clouds with Differentiable Simulation and Rendering of Deformable Objects