Parrot Mambo simulation unstable over 2 m

I am trying to simulate a flight path using the Parrot Minidrone example given by asbQuadcopterStart. Any trajectory that I set that goes above 2 m causes the drone to go unstable (rolls or pitches back and forth until a model assertion terminates the simulation). I have narrowed this down to either the state estimation using optical flow, or the generation of the optical flow data in the sensor simulation.

If I try to disable the optical flow above a certain height, there is no more XY position tracking.

Has anyone seen this issue before? Is there a simple way to fix the optical flow algorithm?