More of a question for Roachey, but from what I understand we are using keyframes. For anyone who doesn't know what a keyframe is (I'll prob miss something here)... we first create a blank animation with the default stance of the character and move their "bones" to where we require them to be and then add a keyframe. Each keyframe is like a little series of movements when completed makes the character look like they are doing some sort of action or movement. With a little practice you can make movement fluent doing away with the need of a MO-CAP suit. (Motion capture suit.)
Physics are rag doll effects, gravity, character/AI hitting something and knocking it over or moving it to another location etc.