I've been researching drone and landing hazard detection systems. The publicly available information (suspect the military is further advanced) is pretty encouraging. My take is that Dragonfly could use any of several systems (and might use a combination) to select safe landing spots. Several of the systems are for landing rovers, etc. on Mars, where there's little time for getting to the right solution. Dragonfly presumably would, once it arrived in its general target area, have plenty of time to scan the area and select safe landing spots. The advantages of a drone with a large battery and thick atmosphere!
I wonder if part of the plan would be arrive at an area of interest, map it in detail, and auto select a safe spot for the initial landing. Then human operators could select specific local spots to fly to and land for follow up studies before heading for the next more distant area of interest.