One more thing - the idea of shaping behavior through successive approximation which sea mammal trainers know so well is such an important tool for police dog trainers - but few I have encountered realize how to employ it correctly.
For example, typical issues that dogs flunk out of class for such as environmental issues with tight spaces, slick floors, open stairs, etc. can usually (not always) be solved by using successive approximation with appropriate rewards. Many trainers think, incorrectly, that this method takes too much time,but if done well and thoughtfully, it actually works relatively quickly. If this is the approach taken from the beginning you also avoid creating problems that require fixing later.
This works so well when you are in the teaching phase of training obedience, agility, building searches, you name it!