1 paper · 1 filter
Jonathan C. Kao, Jason Chan, Andy Wang
We introduce flow control of vision-language-action (VLA) models, a simple and effective way to steer VLA actions in real time through generic inputs, such as a keyboard. This meth…