· Contributors · Organizations · Search
ASE: Large-scale Reusable Adversarial Skill Embeddings for Physically Simulated Characters
Research & Education
Full Conference Supporter
Virtual Conference Supporter
Exhibitor Additional Full Conference
Exhibitor Full Conference
DescriptionWe present a large-scale, data-driven framework for learning versatile and reusable skill embeddings for physically simulated characters. The skill embeddings can be trained using large unstructured motion datasets, leading to a diverse repertoire of life-life behaviors. Once trained, the skills can be used to synthesize naturalistic motions for new tasks.