An interesting question about how much latency is acceptable for UAV operation. http://lnkd.in/dx-BfVk  My answer, based partly on my radio control flying experience, is that it depends heavily on the context. 200 milliseconds is too long for stunt flying, but not a problem for flying larger UAVs at higher altitudes. The operator has to “dial in” their reflexes to the situation, just as sailors do with different sizes of sailboats.  Here’s an example where low latency is essential:  

A seeming paradox is that longer latencies are acceptable only at higher Stages of Control. (See my draft book for discussion of this concept.)  At the high end (what I call Computer Integrated Flying), if enough knowledge is embodied in the aircraft, the operator can pull back entirely from flying, and switch to “commanding” the aircraft.

Here’s my full comment, on the group discussion:

Back to the original question. I think part of the confusion is that the acceptable lag depends heavily on the context.
A lag of 150 to 200ms sounds to me quite long for human operators to feel no effects. But my context is very different: flying RC models maneuvering at low altitude, using either line-of-sight or FPV (first person video). The general wisdom is that even 100ms lag is enough to mess up feedback, and therefore reduces control. Trying to do “pylon racing” with 200ms response would be very difficult. Trying to do acrobatics with an RC helicopter might be impossible.
So the nature of the activity matters. The other factor is the operator’s ability and “set point”. relative to the mission. This is visible also in the type of airplane. Some small RC airplanes (<1.5 meter wingspan, speeds > 150 kph) need very fast feedback to control properly. But others have physical reaction times that are already > 100 ms; they have small control surfaces, large angular inertia, and comparatively loose physical control linkages. These larger planes are quite feasible to fly, but the operator has to get into a different rhythm: make a smooth gradual control action, wait for a response, then adjust if necessary. (It’s just like sailing: 5m vs 10m boats respond to the helm very differently. I can sail both without difficulty, but it’s a very different experience and it takes a few minutes to get accustomed to the “feel” of a particular boat.)
Clearly, indirect flying (i.e. programming a new waypoint) can accept much longer latency. However (at least as of 2 years ago), operators flying large military UAVs like the Predator could not land the aircraft remotely; they had to hand off the landing to someone at the local airbase. The round trip time via satellite was too long for safe landings, e.g. in case of a gust of wind. Once the UAV is programmed with its own closed-loop high-speed control for landing, latency is no longer an issue.