Texas Hold’em Poker Bot

AI Quick Summary

Texas Hold 'em is one of the most popular variants of the card game of poker. In this Bachelor project, the student will develop anArtificial Intelligence (AI) agent to play poker in a real setting. The poker Bot should have a data capture mechanism for quickly capturing the table play: this could be a camera, using computer vision, or it could be a keyboard entry system by the assisted player. Initial situation and problem definitionThe poker game of Texas Holdem is a highly popular card-game of gambling and strategy where players have incomplete information. Despite the simplicity of its rules, it offers extremely rich scenarios of play. Strategic thinking, luck, bluffing emotions and behavioural ticks all play a role. Accordingly, it has ben a difficult game for computers to play well, in contrast to games of complete information such as chess. That was until 2019, when researchers from Carnegie Mellon created the first AI poker bot to beat professionals at a multiplayer game [1]. The problem is that it is not straightforward to apply this technology in casual play between friends.Aim of the work and expected resultsThe goal of the work is to develop an Artificial Intelligence (AI) agent to play poker easily in a real casual setting. The poker Bot should have a data capture mechanism for quickly capturing the table play: this could be a camera, using computer vision, or it could be a keyboard entry system by the assisted player. Once data is captured, the agent should be able to draw on a betting strategy to optimize its utility, i.e. to make money over many rounds of play.Desired methods, procedureA key element of this project is the development of strategic play in this context, using an AI agent based on game theory, or reinforcement learning, or some other similar approach. As a final demo of this project, the student will participate in a real poker game with the assistance of the poker bot. The student is free to use poker libraries and open-source solutions for game play as part of this work. Or to try and develop their own solution from scratch. In either case, it is not sufficient to just come up with winning play, There should be some description and analysis of the learned strategies, so that we can gain insights into what constitutes good game play in poker, and to see indeed, if the agent can be creative inventing new winning strategies.Creativity, variations, innovationThere is huge scope for innovation here. Both in terms of the algorithmic piece (improved game play) and in terms of the practical piece (real game play). There is also the room for stretch goals such as automated data capture e.g. using a computer vision rig.

Project Idea Metadata

Project Idea Name: Texas Hold’em Poker Bot
Date: 8/16/2022 1:12:07 PM
Administrators:
- Transfer Services I

Project Idea Description

Initial situation and problem definition

The poker game of Texas Holdem is a highly popular card-game of gambling and strategy where players have incomplete information. Despite the simplicity of its rules, it offers extremely rich scenarios of play. Strategic thinking, luck, bluffing emotions and behavioural ticks all play a role. Accordingly, it has ben a difficult game for computers to play well, in contrast to games of complete information such as chess. That was until 2019, when researchers from Carnegie Mellon created the first AI poker bot to beat professionals at a multiplayer game [1]. The problem is that it is not straightforward to apply this technology in casual play between friends.

Aim of the work and expected results

The goal of the work is to develop an Artificial Intelligence (AI) agent to play poker easily in a real casual setting. The poker Bot should have a data capture mechanism for quickly capturing the table play: this could be a camera, using computer vision, or it could be a keyboard entry system by the assisted player. Once data is captured, the agent should be able to draw on a betting strategy to optimize its utility, i.e. to make money over many rounds of play.

Desired methods, procedure

A key element of this project is the development of strategic play in this context, using an AI agent based on game theory, or reinforcement learning, or some other similar approach. As a final demo of this project, the student will participate in a real poker game with the assistance of the poker bot. The student is free to use poker libraries and open-source solutions for game play as part of this work. Or to try and develop their own solution from scratch. In either case, it is not sufficient to just come up with winning play, There should be some description and analysis of the learned strategies, so that we can gain insights into what constitutes good game play in poker, and to see indeed, if the agent can be creative inventing new winning strategies.

Creativity, variations, innovation

There is huge scope for innovation here. Both in terms of the algorithmic piece (improved game play) and in terms of the practical piece (real game play). There is also the room for stretch goals such as automated data capture e.g. using a computer vision rig.

Texas Hold 'em is one of the most popular variants of the card game of poker. In this Bachelor project, the student will develop an

Artificial Intelligence (AI) agent to play poker in a real setting. The poker Bot should have a data capture mechanism for quickly capturing the table play: this could be a camera, using computer vision, or it could be a keyboard entry system by the assisted player.