This member has provided no bio about themself...

Report RSS Experiments

Posted by on

I spent a couple of hours trying to see if I can figure out a way to make the AI pack up an mcv and use it for expanding.

Unfortunately it all ended in failure. I hardly got any results from it all, there was actually not really anything to tell if I was doing it right or not.

So I can't call this a good effort, I did try hard though.

I did however change up how the AI use engineers for capturing oil derricks and such.

It should be a lot better than what it was. But I still have got lots of testing to do it would seem.

One thing I've begun to notice is that my new setup is even more similar to EA's original look.

What I'm testing now, is combining a lot of new expansion times and styles into both Hard & Brutal.
From conclusion of replays I've seen of players fighting my AI, I was wrong to make the AI focused on teching and expanding faster. They just get overrun too quick by human players.

I've got a lot of stuff done at the best it's going to get, but I can still improve the AI strategically.
So I am going to keep creating new things and test them like crazy! I'm thinking on making an infantry manager to stop AI from stupidly using infantry only armies.

Cause there's a minimum units per army needed for the AI's attack.

And infantry contribute to that number, by making infantry into their own manager I can make an AI only consider vehicles as their army. Which also means I can lower the minimum needed to launch an attack, Infantry managers will normally want to stay next to the nearest vehicles.

This sounds okay, not great.. but even so, it would also be hard to pull off.

On the bright side I can make the infantry armies have better targets. They can focus on killing other infantry units and anti infantry units.

But that's just one example of what I can try and test.

I'm also going to see what I can do to improve how the AI uses cryocopter forces.
As of 1.01, the manager can still do a dumb mass cryo's in base mistake, never using them at all.
Which is just simply bad and a waste.

My personality I'm currently using for tests is named "Ace", it's going to be the new personality I'll release with 1.02. Ace is going to be using "pro" standard build orders, and will more or less try to imitate real human strategies used on the pro level of ra3.

The Ace AI is going to be something experimental, when I release the update the Ace AI will be the only AI using many of my new additions to my AI.

Of course I won't be neglecting all the original personalities, any important breakthroughs of strategy I find with Ace will be implemented in all the AI.

Ace is just going to be the AI which will use experimental ideas, these idea's include a change of defense preference, sensitivity to nearby enemies and such.

The AI will be more hesitant on expanding for example, with my old AI the brutals need a lot of things to make them decide NOT to tech or expand. But with Ace, it's going to be much more cautious about such things.

Anyways, don't expect this update soon. I'm just going to be doing it, little by little.

Post a comment

Your comment will be anonymous unless you join the community. Or sign in with your social account: