How would you design a fallback mechanism for an autonomous system when the primary behavior planner encounters an unknown state?
