Thread: reborn and jv-bots ?

Sor bots? DO WANT BAD +10000000000

Originally Posted by own3mall

Sor bots? DO WANT BAD +10000000000

*likes* :P

Can someone upload version 3? I want to take a look and see if I can modify / possibly fix it. Sor's working on a framework (looks at Sor... pokes with stick). It's an awesome idea, but we need smarts bots and auto-downloading in this game first.

Version 3 is a Proof of Concept. It will take you quite some time to create something that rivals version 1 with it.
zzzzzzzbots3.pk3

Like I said before, my priority now is the Framework. I'm hoping to finish it this summer so it'll be released along with the client patch.
After that, my last project will be the bots. Personally, I think that's how it should be. We should first release something cool to boost
the online community and then release something cool that people can play client-side out of nostalgia for years to come.

Of course, the framework development would speed up if someone were to aid me. I'll have to clean up the code first; before I had gotten
sick last week, I was doing a little overhaul in the implementation that's only half done so I can't load up the framework right now.

Anyway, try to wrap your head around what jv's doing. While you'll likely won't have to implement something as complicated as the animation logic
there (provided you base your project on this mod), the overall complexity will still be quite high. That is, how smart you want the bots to be and
how feature-rich and optimized you want the mod to be, will determine how long it'll take you and how hard it'll be.

In particular, the pathfinding algorithm will be a pain to implement efficiently; even disregarding the algorithm, just the thought of linearly
searching all nodes for the 'closest one' will cripple your mod on larger maps. But it's definitely feasible to design a good, speedy pathfinding
algorithm (even one derived from A* or dijkstra).

Bottom line: smart design choices followed by an efficient implementation are essential.

Thanks Sor. I'm confused though. I loaded up version 3, and it works perfectly? All I'd have to do is add bot spawn points for each map, add jv's bomb objective handling for obj maps, and it would be done?

I've never tried either version until tonight. Jv's versions is a joke. This one looks pretty sweet... what would need to be done? The logic seems to all be there, and it works?

I'd just have to get reborn to play nicely with it and not think it's a projectile attack... am I missing something here? The logic of the mod itself is a bit over my head with all the vectors and stuff, but I get the general gist of it.

Is there a good explanation somewhere on vectors as they are used in MOHAA? I would assume that multiple vectors are used to return the distance between them in terms of the x, y, and z axis... but I'm probably wrong. Thus, bots could move a certain distance based on the vector differences between the closest enemy / player / bot?

Actually the bots move randomly (you'll find it somewhere in fight.scr I think) which was to test the animation and movement script.
There are no path nodes and no pathfinding so your bots will just move randomly in a circle until there is.

Even the combat is a pure test. The bots receive a weapon and shoot the first bad guy they see. Since both teams are spawned in
front of each other, all of this may give the illusion of a working mod and of 'smart bots' but don't let it fool you.

They are dumber than the single player bots (and that's saying something) because they have no behaviour apart from

Code:

if cansee enemy
      move rand() direction
      do some rand() leaning or jumping
      shoot at enemy

That is because jv was planning to add a secondary form of 'moving' (as seen in the anim.scr script), specially for combat
(and other specialized situations) where the purpose is not getting from A to B but to not get shot by B or to handle obstacles
(dynamic obstacle as well) and the like.

Vectors are used for a lot of things. They describe position and angles as seen in the game; since it is your job to program
your bots to not only move but move through this game world, you will be manipulating a lot of vectors. For example, a neat
trick that jv often uses is calculating the dotproduct of two vectors to check if vec A is 'behind' vec B.

Oh the Joy - was great reading this thread !!! - BOTS are so important - EVEN the current versions are fantastic but without the autodownload the servers dont get players so they just sit their empty but they work great in current format although AI is poor its great fun and I for myself enjoy playing them and it does help you improve your game as they often dont do what you would expect
I understand the paths but they are a great feature so any improvement on BOT feature would be Fantastic!!!

I remember a while back trying to get Elgans zombies stable on multi player - we started to get a populated server - players were downloading the mods and enjoying it and had scope for many game types such as all players v zombies - axis and allies v zombies - few players on either side backed up with extra AI zombies and so on.

JV's Versions were a great effort and yeh easy kills but still great fun

Think we chatted about this a while back - Empty server no players - You want to play on that server but no players - you turn up and wait for 10 mins no extra players comes you leave............

Where as you join an empty server but they have auto bots on!! so when you join insted of you wandering round endless shooting walls some default bots appear to entertain you. so keeps you on the server longer till more players show - the better the bots the longer you stay

Ratios !! - so optional how many bots per human players

1 human player = 5 BOTS
2 human player = 10 BOTS

Depend on the bot skill and game type as in you could do Axis and Allies only 2 human players = 1 human on Axis the other on Allies but the bots get split up so you end up with a team game of 6 v 6 and so on.

For Stock maps I guess this can be done under pathways already on a custom map the bots have to be created using pathways so understand thats a big new issue but sticking with stock maps the concept gives a new server owner many options to promote the game and I have often ventured on to my bookmarked servers to find them empty so the chance that bots will be waiting is a GREAT thing and then YEH once you have many human players on the bots are scaled down and once server full no bots at all

So POC, Multiple Bot Spawnpoints, and several path nodes for initial starting movement (until enemy is detected / player uses the fire key) = finished better working bots? I could possibly do that...

Again, seems to me that most of the logic is already there. Them bots are mean and hard... lol

Spawning pathnodes won't do it, you have to write the entire pathfinding logic yourself.
This is the price that needs to be paid to make the bots fully server-side and adapted to multiplayer gameplay.

In that case, I'd have no idea where to start for the most part. My initial idea was to keep an array list of each bot and player's position vector (updated every 10 seconds), have a bot find the closest player to its current location based on the list, and head on out towards that general direction. The list would get updated whenever a player fires, and the bots would respond. How to get them to not run into stuff and get around the maps without them looking retarded... lol I don't know if I can do that.