![]() Then there are 3 different fields for pushing/knocking the opponent player into and a player being knocked over can be "Knocked Over", "Stunned" or he might be "removed from the board" - this gives us 28 possible outcomes for a simple Block action.Īnother example is a standard player who can move approx. In the first two cases above, the attacking player can choose to "Follow Up" or he can stand where he is. One single standard Block action results in "Opponent pushed back" or "Opponent knocked over and pushed back" or "Opponent knocked over" (the occasions where the attacker or both players are knocked over is irrelevant for the Al's planning because it ends the turn). Which kind of AI gets implemented in BBalc is not yet defined and it will possibly be a mixture of the above called approaches and others. a ST3 versus ST3 block is effectively equal to a ST2(+2 Support) versus ST3(+1 Support) block. This results in abstract patterns to work with, e.g. Based on the rules we can calculate probabilities for every possible player action. In BloodBowl we have the definitions of the board, the players, the teams and the rules to work with. Some prominent approaches to AI are expert systems (using predefined rules to solve a very special tight bounded task), neural networks (these try to simulate the functionality/learning behaviour of the human brain) and genetic algorithms (they learn how to behave in their environment by using evolutionary methods). Another important thing to look at is the AI s ability to learn from its environment and decisions. There are several approaches to artificial intelligence and in fact it is fundamental to define which kind of intelligence behaviour you want to simulate. Each team has his way of playing and game play depends highly on the skills owned by your player. Because of that there is no definite way to BloodBowl. There exist lots of teams, lots of skills and players on the board can act in several ways. ![]() but this article addresses the "ordinary" BloodBowl player. Simultaneous I started the BBalc Wiki, in order to provide documentation about my work, and to offer the occasion to discuss all parts related to a BloodBowl AI.Īnd this is what this article is about: I need YOUR help! (In fact I would love to say: WE need your help, but after more than 1000 project page visits on SourceForge and 40+ downloads later, I got exactly 2 sentences of feedback, which can by found on the projects SourceForge project page.) Well, I need support in coding, web design etc. ![]() Meanwhile it became obvious that an BloodBowl AI would be a very huge pile of work and that this task cannot be done by one single human, at least not in the next 10 years.įinally I released my work, now known as "BloodBowl a la carte" (BBalc) under an open source license and put it on. I created a basic framework that knows most parts of the BloodBowl game and some simple analyse functions and a basic GUI to display game data and analyse results. But after playing some games on FUMBBL, I found out that one little detail is missing in every BloodBowl game I know about - a computer player.īeing intuitively fascinated by Artificial Intelligence I decided to fill the last white spot on the BloodBowl map and started to work on an AI. Reinventing the wheel again and again makes no sense for me, so I stopped that project. Few weeks later I found out that there already exists some BloodBowl games. ![]() In order to make the project more fun, i started writing a BloodBowl game (a game which I used to play some years ago.). Some days ago I decided to write a TCP Client/Server Application for learning purposes.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |