Gijs-Jan

Could also be the A.I.: Civilians now run for cover / the least visible places.

This could be partly blamed on me, I think. :-) The problem is that the A.I. now evaluates all tiles surrounding the unit, and (among other criteria) selects the one that is the least visible to use for ambushing / hiding. These tiles are evaluated dynamically, and so use the underlying meta of the tiles. The result is then if you have a glitching / improperly setup tile; you get an A.I. unit that thinks: Damn, I'm near invisible there! Off we go! It still approaches the tile with the game's pathfinding; so if any alien can move there, so can a Xenonaut. You could try standing next to it and force shooting the tile. (CTRL) (Although the logic that differentiates between what blocks a shot and a unit could be different) Another approach could be to have xenonauts approach the tile from 1 side; which could force the A.I. out of hiding if enough are present.

I only worked on pathfinding so far, and haven't yet looked at the targeting. It probably goes through the list of all targets, and (from a code perspective) selects the one it can fire at first. This build could contain some enhancements with regards to FPS. I don't know for sure if they're integrated yet, and if not; next build could see an even higher FPS.

Do you mean a smooth fps?

Hehehe, yeah, it's a problem of the map meta settings. The A.I. evaluates all the locations it can move to; to summarize, a location which turns the unit invisible is pretty valuable :-P

Is this during ground combat; especially during the A.I.'s turn? It's a bug I found out about only yesterday. I'm currently working on getting it fixed.

C++; although I would have really preferred almost any other language :-P Anyhoo; do any of you guys have a savegame which triggers this bug? I might be able to gander a look at it in my downtime :-)

Did this happen when entering land combat, or entering the alien turn?

I know your pain, and staying up will just result in the next day being wasted. :-)

I have but one personal goal in this project: Fear the Reaper..

I will. :-) Its behavior Chris requested; so I do hope so :-) As for individuality; it depends on how far I get with development. The current system in place is entirely built around simple unit based reactive behavior; I'll have to start from scratch as the paradigm is just the complete opposite. I just uploaded some basic augmentations to the AI for pathfinding. (civilian fleeing, basic cover usage) I missed the deadline for the weekend alpha as I spent most of the night dealing with BSOD (driver iss.); sorry guys! Ah the great joy that is developing on Windows... *sigh* :-p Anyway; don't expect too much, as I'm mostly now trying to figure out how to work with the old code: what to keep, and not. But at least it should be better than the random movement ;-) Ow and Buddhist civilian is still possible. The AI only cares about where it wants to go; not about how its going to get there.. :-p

This would fall under the category "learning". The problem is that you separate this into offline and online learning. Offline learning would mean analyzing (thousands) of plays; and adding this data as knowledge to the A.I. Online learning would mean learning while the player is playing; which is very, very difficulty as so many factors change in this game from map to map. I.e.: How would I know where the player is coming from, if the map is randomized between each battle?

Sorry, still running on the old true and tested model Nexus-5 series. Pretty Relevant

Welcome right back at ya! :-) Very, very good question. This is the base difference between planning and reactive A.I.; as for this situation the A.I. would have to analyze the different outcomes of the combat. (Or we could make it guesstimate) I'm not entirely sure whether I will be able to finish it by 9th of Oct; as the remains of the last few A.I. coders (also very talented, but they wanted to go a different way), keep hindering me a bit. But it's a definite goal to have it in there before I call this thing "done". :-)

Shouldn't be any problem at all :-) The A.I. should pick options and decide on the best course of action, given some randomness.

No prob. Do you mean combat-wise? I can't really think of examples that spring out; although I know people that designed (learning) combat A.I.'s for Neverwinter Nights. The problem with game A.I. is that the game industry has become very, very good at creating "good enough" A.I. (An almost direct quote of Alex from GameAIDev.com). There's not much incentive to construct A.I. for the hordes of goblins that will only be on screen for a couple of seconds, or A.I. that is practically invincible. :-) In other terms, there are games like Dwarf Fortress where aspects of "A.I." (I'm using the term extremely liberal now) really contribute to the game.

Thresholding function! Or just plain assign a slight positive value to your current location.

We saw the system produce stranger things :-P It would learn solely based on statistics, and not really "know" anything on what it was learning. I.e. If A occurred a lot, when B did as well, well, then maybe they were related the system assumed! And that was the day the expert system said breast cancer was caused by a clear blue sky :-/

Funnily enough, our second-year project was an expert system based on Bayesian Networks (A system that gives a prediction using probabilities; I.e.: if the grass is wet, either it probably rained, or you turned the hose on). Which was a 2-phase system: first the system would "learn" to recognize the disease by it's symptoms. Then it would be tasked to diagnose patients. You see these systems employed on the internet nowadays. Fun fact: even though they outperform doctors in diagnostics by far (in terms of error rate); you won't see them employed any day soon. People just don't trust a computer with that kind of task :-)

Congrats on making the game! Also, for games of that magnitude I would really recommend Monte-Carlo Tree Search. It's well suited for AI's of this task, as it's prime focus is large domains where the eventual outcome of a strategy is much more important than the intermediate steps. (I.e.: nothing might really happen for 300 turns, but the end-game difference is huge) If you need any help or tips on it; just PM me. As for the atmosphere: Else I wouldn't have joined the team :-) Will the A.I. adapt to a tactical situation like you described? It should. Would it learn based on your performance; i.e. would it make a different choice when presented with the exact same situation twice? For the 9th of October? No, sorry. However; I'm with the project at least up until January, almost fulltime. And it IS something I (very) briefly discussed with Chris, and it is something I have constructed before. (For Battlecode ) If everything goes as planned, it shouldn't be to difficult to make the AI behave differently on different difficulty levels (3x diff combo-breaker!).

Ugh, script FSM's, I really abhor them. It's nothing more than just writing a script of what the A.I. exactly should do under which circumstance. There wouldn't be much of a challenge in writing them in this style; just, like you said, an insane amount of paper work. I prefer to use the FSM to set the A.I. in a certain behavior pattern and let it figure out itself what it should do instead of dictating it :-) I'll try to explain more clearly in the coming posts! Ah, you caught me! Yeah, I was/am a big fan of both series (and for both series, I still prefer the original); the atmosphere in these games is still something I've not yet seen matched in any other. Simply put: it won't. I want to be completely open and honest in this, so in detail: All of the decisions it will be allowed to make will only be those it would have made given sight/etc of its units. If you see the A.I. cheating or doing stuff based on something it shouldn't have known about; then I'm doing something wrong. To explain this a bit; I have to make some shortcuts here and there as belief systems (Knowledge based on uncertainty) are very costly in general. So behind the scenes, some trickery might occur.

Yes!

I am designing everything with modability in mind. However with A.I. it's rather tough to have both "intelligent-decision" making AND complete control over what the A.I. can do. In other words: you'll need to trust the A.I. to make the right decision. In essence; mods will allow the enemy units to change behavior (P.e. Rambo-style <-> Cover-sneaky-ninja-style) quite easily, but they probably won't be able to exactly tell them what to do. Can't have civies becoming too smart! But maybe I'll allow the randomization to produce some that are smart enough to run to the dropship. :-) (If they can last long enough to get there that is..)

Thanks for the warm welcome! In general sense, this is true. Rather simplified: in realtime games (from a NPC perspective), the A.I. is mostly reactive as its calculation time can be measured in milliseconds, if not less. Turn-based games however, generally allow (and require) the A.I. seconds of calculation time as we hope to beat the player in planning, not in reaction-time. It does depend on perspective though: in realtime games we could have an extra overarching A.I. that takes its time to calculate correct squad positions. The whole problem is that in realtime, the A.I. constantly needs to keep up with new information, in a continuous environment. In turn-based games, the A.I. is spoonfed events and can take its time analyzing them. Definitely a major point, and one I am taking my time to get right. Next couple of journals should focus on this. Don't expect anything big soon; it's a big undertaking and I'm trying to do more than several months of work in a very short time. :-)