I can personally endorse forms of the answers provided so far.  

 1. Work on more than one problem/project.  Open problems are hard.  Instead of chasing after one fish that could be out past the horizon, cast a net or set out several poles.  Or, if it is important to stay focused on one project, work on more than one piece of it at once. 
 2. Make your work a one-person seminar and take notes.  I have always kept pages of calculations when working on problems.  At some point, actually in mid-career, I realized that I was not just writing down calculations, but also reasoning and Q&A.  Thinking often amounts to "the voice inside your head".  (This may say something about the nature of schizophrenia; it seems that we all hear voices.)  I ask my PhD students to write notes about what they are doing, not mainly for me to read them, but as a research aid for them.
 3. Give talks, ask questions, and answer other people's questions, again just to keep moving.  This site, which is among other things a reincarnation of the old sci.math and sci.math.research, can be very useful for this purpose!