Recognising, from a camera feed, that a physical step has been completed, to auto-advance the displayed procedure step.
Vision-capable model checking the current camera view against the expected post-step state.
Vision-capable model API