Xi-Cheng Gong

A simulation flywheel recursively bootstraps embodied data, generating richer scenes and tasks for robot pretraining and evaluation.

A unified architecture achieves SOTA embodied AI by jointly learning physical reasoning, world generation, and action prediction, outperforming prior models in generalization and long-horizon tasks.

Existing embodied QA benchmarks focus on narrow reasoning skills; EQA-Decision finally offers a unified, large-scale dataset spanning static scene understanding, spatial reasoning, task dynamics, and instant decision-making.

LLMs can dynamically replan sub-goals during embodied instruction following, leading to state-of-the-art performance on ALFRED.