SwitchVLA: Execution-Aware Task Switching for Vision-Language-Action Models Paper • 2506.03574 • Published Jun 4, 2025 • 1