Excellent advice, often ignored. To your comment, I would add that you shouldn't limit your problem to something small or "doable." List and describe as grandiose a problem as interests you, then carve out a piece of that problem whose scope is appropriate for your project.