A robot's perception node detects obstacles and publishes them as a list. A planning node subscribes and computes a collision-free path. If the perception node crashes, what happens to the planning node?
AThe planning node immediately crashes as well, because it depends on perception data
BThe planning node continues running, using the last perception message it received. If perception doesn't recover quickly, the planner operates on stale data, potentially creating hazards
CROS automatically restarts the perception node and resumes publishing
DThe planning node switches to a hardcoded fallback algorithm without perception data
This is a classic architectural evolution: ROS 1 was elegant for its time (peer-to-peer via a master) but had limitations that prevented deployment in real-time, fault-tolerant, and wide-area scenarios. ROS 2 solved these by adopting a more mature standard (DDS) and accepting additional complexity. The lesson: architectural choices that are fine at scale N become problematic at scale 10N.