1 paper
Chaeyeon Jung, Juyoun Park
Vision-language-action (VLA) policies are expected to operate robustly across variations in the robot's initial configuration, yet aggregate task success can conceal pose-specific…