Hey
MEMS are related to the IRU part. They probably have a potential to cut prices. IRUs and GPS are very complementary to each other: IRUs have a drift tendancy (summing of errors on acceleration increments) while GPS will not drift but will have larger instantaneous errors. They can correct each others. With 3 GPS antennas, the GPS can also estimate the attitude angles, enabling a deeper hybridation with the IRUs. A complete hybridation (we call it sensors hybridation in French but in English, it might be termed as sensor data fusion ?) may be possible with all the air references, including new sensors like laser anemometry (which can remotly sense the 3D airspeed field, hence also mesure angles of incidence, slide, etc...) or a Lidar (new weather radar for improved detection of turbulences/CAT, with doppler/interferometry capabilities).
I may be dreaming, but this full hybridation of IRUs, Air references, Laser Velocimeter or Lidar, barometric/radioaltimetry altitude, and possibly other air reference sensors (cinetic pressure transducers on the airframe or in the compression stages of the engines as suggested by others), implemented via a large extended Kalman model (mainly describing the airframe & flight/fluid mechanics, the sensors and their measures... the underlying physics) would provide an optimal error detection/isolation scheme (nothing to do with constant thresholds in magnitude/duration to detect faults as it is done) and the capability to estimate some failed air parameters via all the others and through the real time monitoring the aerodynamical authority of the airframe (and its control surfaces, via microorders added to the normal orders). This real time aerodynamical authority estimation (through slight modification of the "posture" of the aircraft) coupled with the direct sensing of the air stream and the resulting linear/rotational accelerations provided by the IRUs (and possibly laser/lidar imagery of the air stream) would be close to the proprioceptive sense used in bird flight. This extended Kalman model would "know" that spikes of a few degrees magnitude and of a few milliseconds duration on the AoA or pitch are not a real thing in flight mechanics it has to react to (fault detection & exclusion), all the more if other paramaters remain unchanged (vertical acceleration). It would also know that a CAS drop approaching 200 kts in amplitude in a blink of an eye is not possible if the control surface authority and other parameters remain constant. It would exclude corrupted measures in an optimal manner, relying on measures consistant with each others from an aerodynamical point of view. Otto would be a lot smarter ?
This sensor hybridation scheme is used by some military flying vehicles.
Nothing to do with the air reference improvement but in difficult conditions, the reading of the instruments may be a problem according to some pilots: the military have head up displays, why not for the civilians too ?
Jeff