Zhi-Yuan Li

This work proposes StageGuard, an agentic distillation framework for accurate and efficient stage-transition decisions, which combines teacher-model reasoning with demonstration trajectories to generate structured explanations of subtask completion and policy switching.

Key contribution not extracted.

Current VLA benchmarks may be overstating real-world readiness, as models succeeding by standard metrics often exhibit unsafe behaviors and poor robustness.

Web-scale video pretraining lets robots handle real-world chaos better than vision-language models trained on curated robotics datasets.