Okay then, you seem to be right on top of most of that, lets focus on autoguiding.
If you have a image comming in and some software that can scan that image in real time then you could choose a bright star in the field of view and "lock" its position on the screen. By locking its position everything else on the screen would be locked too for a reasonably properly aligned mount!
The way this works is the software views the CCD chip like a chess board and the stars like chess pieces. You have just told it to say lock it so that the Queen (a bright star) is locked on QB-4 (Queen's Bishop - four steps forward of the base line) a known pixel on the CCDs grid.
Now a star could wander off position based on either recurring errors (bad polar alignment or imperfections in the drive gears) or intermittent errors - knocks to the mount or OTA, vibrations such as wind gusts. If the errors are regular - pointing alignment or drive imperfections - one calls them periodic. In one revolution of the mounts gears you are delta x and delta y out of position - then this can be observed and corrected. That function is called PEC - periodic error correction. But to fix either random or periodic errors you can use auto-guiding.
So from the above example for whatever reason your selected guide star moves off your selected pixel. Then the software distinguishes this and generates the movement commands to the mount that should move this star back onto this pixel with a set amount of alacrity (the aggressiveness of your corrections). These commands have to be sent from the PC - say via your serial port - as an electrical signal - a usually a 6 wire flat cable - into an auto guide port on your mount. This port can make fine tuning instructions to your mount; like move three squares up and one square left on that chess board of our example.
Now apparently there is no universal standard for auto guiders - commands, electronics and signalling. But there is a defacto one because the most popular and therefore prevalant brand and model of autoguider was the $1,500 SBIG ST-4 model (which I believe has now been retired). So many manufacturers design their mounts to be compliant with the way this specific model of auto-guider worked. e.g.
http://www.astrocruise.com/st4tips.htm
Now a computer program can emulate the way a dedicated piece of technology can work. A Meade DSI or LPI CCD signal can therefore be interogated (say by AutoStar Suite) a bright star selected and locked, and the software issue commands - to a Meade mount - to keep that stars position locked.
Now again Meade and Celestron don't follow universal standards to drive their mounts (pain in the posterior this lack of standards). But again you can emulate one mounts control commands and translate commands for the Meade into commands for the Celestron. So I use Meade imaging gear and software and run it into a Meade -> Celestron command convertor and send it into a Celestron CG-5 mount.