I suspect you could probably implement some kind of stabilization system, however I think you would have to jump through some hurdles to get there (eg using external sensors + implementing some kind of fail safe mechanism) that it's viability is questionable in terms of effort and time. In my opinion the opportunity for this board would be to supplement a flight controller by off loading non critical tasks to it or it forming part of the group station set up.
That would strip down Udoo Neo from the purpose of having M4 and internal sensors and obsolete them in favor of faster external sensors and safer M4 controller performing real-time tasks independently. However A9 core in the Udoo Neo is not a miracle when it comes to computational power, and for the Linux co-processor I would probably pick more powerful board of similar size, like ODroid. That can probably handle more computationally demanding tasks.
