TACO: TActile World Model as a Self-COrrector forScalable VLA Post-Training figure
AlphaXiv 中文概览(可滚动查看)