ResearchTasksClosed-loop VLA OOD GeneralizationClosed-loop VLA OOD GeneralizationFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedEmbodiedGen-generated scenes (OOD)EmbodiedGen V2 (N=50)77.9OOD Success Rate2Jul 9, 2026