Maksim Bobrin

Achieve robust text-to-motion generation for long, compositional prompts by planning in the latent space of pre-trained behavioral foundation models, rather than relying on heavy end-to-end motion generators.