Sheng-Ju Qian

Robotic world models can now learn from diverse, yet confident, alternative futures, achieving state-of-the-art prediction quality and extending predictive horizons.

Spatio-temporal masking during student self-rollouts breaks the reverse-KL mode-collapse trap in autoregressive video diffusion, eliminating visual over-smoothing without relying on real video data.

By harmonizing diverse video datasets and architectures, SolarWM enables real-time interactions in video world models, pushing the boundaries of long-horizon inference.