Design And Implementation Of Artificial Intelligence GuanDan Competition Platform

Since Alpha Go defeated Li Shishi,the human-computer confrontation event has ignited the upsurge of artificial intelligence and human confrontation,and artificial intelligence has received more and more attention in the field of intelligent games.Card games have aroused widespread interest among artificial intelligence researchers because of their simple rules and incomplete information game.The development of an artificial intelligence card game platform that can provide efficient optional action sets and low access costs has become an important issue.research content.This article studies the "Guan Dan" artificial intelligence battle platform,which provides a convenient training environment for researchers interested in Jangdan games;provides low-cost algorithm access for developers who are interested in Jangdan games;Top human players who are interested in artificial intelligence battles provide user-friendly clients and,based on the platform implemented in this article,hold the "first " Hongze Lake Cup " Artificial Intelligence Guan Dan Competition".How a platform can widely support the access of most programming languages while being able to communicate with algorithm programs efficiently is still a challenging problem.Based on the above background,this article starts from the actual event background,and the specific work is as follows:(1)Propose and implement an automatic card generation module to efficiently generate optional action sets.Perform index coding on actions in the action space,and reduce time overhead by judging whether the index is in a legal range.The card point list algorithm,the single-card type generation algorithm and the continuous type generation algorithm are proposed.The three algorithms form the card type automatic generation module,which can automatically generate a set of optional actions for the current hand and game scene.The method of loop nesting to extract card actions is time comparison.(2)Realize the language independence of the platform through JSON based on Web Socket.By designing the JSON format in specific scenarios and communicating with intelligent algorithm programs written in different programming languages through Web Socket,language independence is achieved,and a package capable of JSON automatic parsing is provided,which reduces the cost of intelligent algorithm program access to the platform.(3)The GUI toolkit based on Py Qt5,from the perspective of friendly front-end and humancomputer interaction,realize the client and game playback tool that can compete with the intelligent algorithm program.Starting from the actual needs of real players,realize the client with card management function.Analyze the necessity of game playback from the perspectives of developers,real players,and hosting events,design the recording format of the data file,and realize the reading and playback of historical game data.It also describes the meaning of the game replay tool in conjunction with the actual problems that occurred in the competition.