AI (Pathfinding)

Here's the newest update to my game (Skylock: Cyborga). This update has the first bit of AI, in the form of pathfinding.

From the main menu choose the button that reads "Test" to load a simple test level. From the game screen you can cycle through the current trooper by clicking on the "Next" button located in the bottom left of the screen. When you cycle to the next trooper it will determine his allowed destination hex's (highlighted with a light shadow).

If you click on a valid destination hex it will determine the quickest path and plot it (designated with red stars; pre-alpha place holder art). If you click again on one of his allowed destination hex's it will recalculate his path.

It is currently a "dumb" AI, in that if there was a person (or in the actual final game, a wall, barrell, etc.) it would just walk through them. Over the next few days I hope to upgrade the pathfinding to a smarter AI to generate walk-around paths when the quickest path is blocked.

Let me know what you think. :D

if you are interested in shortest path algo's look up dijkstra's algorithm (i cant spell). or look up 'greedy algorithms' in general. they are great for path finding.

Try GameTutorials.com. They've got some AI demos.

Here's the latest version with a "smarter" AI. Of course, it could still use some tweaking but at least now they walk around other troopers, etc.

And the newest version complete with walking. The AI is smarter (though not exactly as 'smart' as I want it). This updated version has the troopers and bots actually walking with their respective movement animations. I do need to tweak the walking anims a little (ie: the speed between the frames, etc.).

thats lookin pretty good. you should have some way of identifying which player's move it is. i never know which guy is about to do the walking.

So jdinger, which path algorithm did you use?

The algorithm is one that I worked out over the last week or so. I don't know that it's patterned after any specific algorithm. I've never studied pathfinding AI.

Basically I start with 2 variables: the Origin point and the Destination point. Then I iterate until the Origin point is equal to the Destination point taking these steps to get there:

1-For each iteration determine where the destination point is from the current origin point (southwest, west, northwest, northeast, east, southeast).
2-Based on the direction given from the first function I populate an 6 element integer array with the preferred order of directions to make the move.
(ie: if I need to move east then the preferred order to try would be:
0-east
1-northeast
2-southeast
3-northwest
4-southwest
5-west
3-Next loop through the preferred order and test if the hex that the current direction would move you to is clear, if so then move to it and break out of the loop.
4-Update the origin point to be the point you just moved to and test to see if the new origin point equals the destination point. If so then you're done moving. If not then start back at step 1.

Quote:

---

thats lookin pretty good. you should have some way of identifying which player's move it is. i never know which guy is about to do the walking.

---

That's in the works. This is very early in the AI. I was thinking of going with a highlight to the hex that the current trooper stands on. Like a mellow flashing red or green or something.