Movement AI needs work
Maybe this has been discussed before, but it bears mentioning if it hasn't: I can kick the computer's butt with a simple player movement algorithm. This algorithm should be considered for inclusion into the next generation of Strategic movement AI.
It goes like this:
1. Examine enemy ship speeds and calculate closest approach.
2. Examine enemy weapon ranges and calculate nearest safe distance.
3. Approach to just outside nearest safe distance, then close to attack next turn.
As I said before, this tactic has won me most of my tactical combat scenarios against the computer.
In phase 2, calculating enemy weapon ranges, I take into account the weapons destructive potential. The math is simple: Multiply %chance to hit by damage at range to get ExpectedDamage at range. This measure of enemy damage potential can allow the player/ai to accept a certain minimum damage that won't pass shields, and assess whether the enemy can be destroyed first.
An additional result could be the creation of a new Movement Classification, called Best Damage Ratio. This would be moving to a range that has the highest FriendlyExpectedDamage/EnemyExpectedDamage, to be applied during phase 3.
|
Threaded Mode