GRAPE: Generalizing Robot Policy via Preference Alignment

Open in new window