They plan to use some form of "site evaluation" to prevent tipping over:
However, technology developments in the last two
decades, notably the revolution in availability of multi-
rotor drones a made possible by modern compact sensors
and autopilots as well as the development of sensing and
control capabilities for autonomous landing and site
evaluation for planetary landers, made a quadcopter or a
similar vehicle a much more feasible prospect in 2016.
They plan to survey future landing sites before attempting a landing:
This paper has more detail:
Once safe landing on arrival is achieved, the rotorcraft mobility capabilities can be exercised progressively— for example, first making a brief hop for a few seconds within the immediate vicinity of the landing site where the terrain will be known from panoramic and/or descent imaging. Depending on the heterogeneity of the surface (e.g., patches of sand), a small displacement of a few meters or tens of meters may enable the sampling of different materials.
Then, flights of progressively increasing duration, range, and/or height can be made, returning to the original, known-safe, landing site. These flights can assess the performance of various sensors—for example, an initial hop may be made using inertial guidance alone, whereas later flights use optical navigation only after the quality of in-flight imaging and the abundance of suitable landmarks on Titan have been verified.
And about the sensors used:
The vehicle makes a sortie over this zone using its sensors (lidar for terrain roughness, imaging, etc.)
When using the reconnaissance method from the graph above, the data gathered during flight will be downloaded and analysed by the operations team before a new landing site is chosen.
Analysis on the ground of the sensor data confirms one or more safe sites within zone B