Tough one.
Unfortuantely, almost all wireless communications methods are highly subject to intermittant dropouts. No matter what method is used, I like to have slave processors on each side of a link, do most control locally, and buffer commands in and out.
RF solution dropouts can occur because of just about any kind of electrical noise. The ultra-fast rise time on the switching waveform of modern AC drives is a typical culprit (yep, they can hit 2.4 Ghz and above). Other problems can be walkie-talkies, portable phones, interruption of line-of-sight via moving parts, even lags in the wireless transmitter/receiver units frequency hopping.
The first thing to try if you must go this route, is ultra-filter everything. Second would be to use more directional, higher-gain antenna's. On the movable section(s), you might even want to look into an array of antennas. That doesn't mean just putting several in parallel, but properly designing a phasing network to maintain proper feedline impedance. Third might be to have redundant sets of transmitters/receivers, but now we are talking real expensive.
Optical solutions (infrared) are generally more reliable, but not with off the shelf components. The best system I've seen for optical transmission has used a large (about 3' diameter) ring of emitter/detectors facing an identical ring on a winding turret. That was a "roll-your-own" emitter/detector array coupled through signal drivers/conditioners to an off-the-shelf mini-IR coupler.
The simple fact of the matter is, for critical applications, I still prefer to use a wired network, even if it is of a much slower speed. RS-485, DH+, RIO, MODBUS, TIWAY (is that still around?) and the other low-speed standard networks are very reliable through slip-ring or slide-plate couplings.
Good luck.