Yes, that's right. The ostensible idea behind subsumption is dead (because it was wrong). But what subsumption did was open up the possibility of putting the AI into the run-time feedback loop rather than the deliberative planning, and that is what all robotic control architectures do today.