GEVRM: Goal-Expressive Video Generation Model For Robust Visual Manipulation

The classic internal model control (IMC) principle: A closed-loop system with an internal model that includes external input signals can accurately track the reference input and effectively offset the disturbance. We propose GEVRM, a Goal-Expressive Video Generation Model for Robust Visual Manipulation. As shown in this figure, to effectively implement the classic IMC principle in the VLA model, some components of our method are adjusted accordingly.