Presentation

Learning Active Quasistatic Physics-based Models From Data
SessionSummary and Q&A: Learning for Simulation
Authors
Sangeetha Grama Srinivasan
Qisi Wang
Junior Rojas
Gergely Klar
Ladislav Kavan
Eftychios Sifakis
Event Type
Technical Paper
Recordings
This session WILL be recorded.
Interest Areas
Research & Education
Registration Levels
Ultimate Supporter
Ultimate Attendee
Exhibitor Ultimate
TimeThursday, 12 August 20217am - 8am PDT
Location
DescriptionWe design a differentiable quasistatic simulator to learn a low-dimensional control space for active models, reconstruct training poses, generalize to unseen poses, and enable further physics-based edits on the output.
Authors
Sangeetha Grama Srinivasan
University of Wisconsin Madison
Qisi Wang
University of Wisconsin Madison
Junior Rojas
University of Utah
Gergely Klar
Weta Digital
Ladislav Kavan
University of Utah
Eftychios Sifakis
University of Wisconsin Madison
Weta Digital