I received this email describing recently canceled Nikon projects:
Nikon were working on a method of embedding camera motion data, as detected by VRII lens data, in a video file. The idea was to record .nev video at resolutions higher than standard (upwards of 500 pixels on all 4 sides) and remove motion shifts by extrapolating the VRII data. This was out-of-camea processing as it was very CPU intensive. The results worked quite well for small, high-frequency motion, but ultimately was deemed not worth the effort. This required significant architecture change to camera internals and was several years from deployment when cancelled.
Nikon also had a prototype "body" that used a laser range-finder system for highly accurate focus distance calculations. This was to be applied to super-telephoto lenses with new technology that allows them to know their focus distance. The pairing worked very well and allowed for a prism/mirror -less, accurate autofocus. The prototype "body" had a fixed laser (think fixed center-point autofocus). Note that this body was an extreme engineering sample and was more of a box than a camera body. The technology was for demonstration. Unfortunately Nikon deemed it not worth the effort to work out response-time and tracking issues. This project was cancelled because of recent, significant progress in phase-detection autofocus. The new phase-detection is, subjectively, as fast as the E-P3 seems to be. It is implemented in all 3 DSLRs that will be released this year. I don't think it will be in the new mirrorless Coolpix Pro models as it's processor intensive and the Pros don't have (that) much onboard power.
One of the expected improvements in the next full frame camera from Nikon is faster AF. Some websites compared the speed of the new Olympus E-P3 to be identical to current models DSLRs. Here is an example of how fast the Olympus E-P3 AF is in low light (there are few more AF tests in the same post):