Tonghua Su

Achieve nuanced control over portrait animations—even for subtle states like thinking or drowsiness—by using hierarchical agent planning to translate high-level labels into precise eye movements.

GaLa's hypergraph representation reveals hidden semantic relationships in multimodal data, leading to a dramatic boost in procedural planning accuracy.

By disentangling structure and motion in the latent space, CoWVLA achieves superior visuomotor learning compared to standard world-model and latent-action approaches.