PCs will merge with consoles (particularly things like Kinect), and this to produce full immersion virtual reality chambers. Think I'm crazy? All the tech required is either here, or less than 5 years away.
That's not full immersion... That's just a big television. Increasing the size of your TV does not full immersion make.
Here's the issue when it comes to VR:
1. For it to be ubiquitous, it must be cheap enough for the masses. That means sub-$1000, if not sub-$500 range.
2. It must be compact. If someone has to build/clear out an entire room just to use it, you are doing it wrong; that would never catch on.
To meet both requirements enabling VR to be anything more than the personal projects of obsessive uber-gamers, one must think smaller rather than larger.
The problem there is that ATM, VR/AR glasses are very poor quality. Having talked to people who worked on a capstone project involving AR glasses and having tried out some VR glasses myself, I was anything but impressed. The maximum screen resolution was about what you would expect from a netbook; something like 1024x768 or 800x600. The glasses produced simply enormous amounts of heat, both making your face extremely hot after 20 minutes of use and supposedly their group went through a couple pairs that fried themselves simply from their own heat after being left on for a few hours. They also tended to have very high power draws, resulting in the frying of several PCs' USB ports.
Yes, such systems will improve, and rapidly, but it will be at least 10 years before that sort of VR glasses will be ready for the mainstream. And still, even a VR display does not full immersion make.
There's also audio; which at the present is typically done very poorly, and is quite overlooked in games and in the coursework training those of us going into the industry when compared to effort put into graphics.
One of the other potentially interesting developments over the next decade or so are EEG based input devices such as the NIA or
Emotiv headsets. Unlike input devices such as keyboards, controllers, Wiimotes, ect, they are nonexclusive. Thus they have the potential to enhance the input ability of a user; particularly when it comes to information about the user which would not explicitly be inputted using a standard input device. If you can get even a very basic, general idea about that user's excitement levels or emotions in real time, the range of features one could implement with that data is quite vast. It will probably be at least about 5 years though before it is able to become mainstream, as it is currently gen 1 technology, and still has a few issues to sort out, with electrical interference being one of the big ones.