Automated blackjack dealing system

Design team members: Abhinav Ajgaonkar, Danny Chung, Jason Fong

Supervisors: Justin Eichel

Background

The game of Blackjack is by far the most widely played game at the casino, due to the simplicity of the game play and the publicity that surrounds it with regards to the card counting technique [1]. The result of the increasing popularity in the game is often reflected by high wages paid to Blackjack dealers and more players wishing to become the next Blackjack guru. From the casino’s perspective, mistakes made by dealers can be very costly and quickly destroy the reputation of the house. From the player’s perspective, Blackjack is a game that is hard to approach for an inexperienced individual, since their decisions during the game play can bring impact to the fate of the other players at the table.
There have been previous attempts in designing systems [2][3] that minimize the need for the dealer to multitask, and hence minimizing the cost attached to the dealer’s mistakes. The patented designs are largely divided into three types: 1) an automated mechanical system that can perform the dealing exercise, 2) an image processing system that can recognize the cards to keep track of the game play, and 3) a completely software-based system responsible for the entire game play. However, many of these designs remain a proof of concept and fail to be commercialized because majority of them can only perform a single task and still require a dealer to be present. Although the software-based systems can completely replace a dealer, there is a lack of incentive due to people’s common belief that software systems are rigged since there are no physical dealings of the cards.

Project description

The main objective of the project is to address all the issues not captured by prior designs by creating a completely automated Blackjack dealing system that will replace the human dealer for the game of Blackjack. The desired outcome of the system is to benefit the casinos, by eliminating the human errors and the ongoing costs from expensive dealer wages and training, while maintaining a similar experience for the game. A secondary objective is to assist inexperienced individuals to become familiar with the game without the need for a human dealer, so that they can improve their skills without investing into casinos.
The system performs all possible tasks required for a complete game play of Blackjack. This includes handling of the cards (both shuffling and dealing), card recognition, keeping track of the states in the game, and interacting with the user through various forms of input and output.

Design methodology

The system is first broken down into three sub-systems, namely mechanical, image recognition, and software and electrical. A design methodology is derived for each system to achieve their individual objectives.
The mechanical system is responsible for the shuffling of the cards, the interaction with the image recognition system, and the delivery of cards to the table.

Flow chart depicting mechanical system process

The shuffling mechanism is derived from a standard card shuffler design, which uses the rotation of motor-driven wheels to push the cards into a storage bin. The delivery and the retrieval of the card to and from the image recognition system are accomplished by using rotating wheels driven by motors, which push the card through by applying a frictional force. The delivery of the card to the table is performed by a shooter, which consists of wheels that can adjust in rotational speed to alter the speed in which the card is shot out.

The image recognition system consists of a camera and an image processing component in Matlab. Image recognition is achieved using a variety of different techniques such as blob detection, which focus on extracting different features of the image. Blob detection identifies areas within an image that is either brighter or darker than its surrounding. The purpose of identifying blobs is to extract features like the area, perimeter, centroid, and diameter of the blobs. For card recognition, the rank of the card is identified by detecting the number of blobs present in the image. The suit of the card is identified by finding the features of the blobs in an image and comparing it with experimental results obtained previously.

Image processing applied to a playing card

It is not necessary to use the general techniques of image processing. Alternate methodologies such as determining the colour ratio within an image are used to separate the spades and the clubs from the diamonds and the hearts. A combination of techniques is applied to create a more robust system with higher accuracy.

The software and electrical system is responsible for the game logic of Blackjack, the interaction with the image recognition system, the automated control over the mechanical system, and the interaction with the player. The core component of the system is a software application which governs all aspects of the system and ensures synchronous operation of the individual processes. The game algorithm determines the flow of the game using basic rules and allowable states in the game. The interaction with the image recognition system is accomplished by communicating with the Matlab process to retrieve the information regarding the recognized cards. The software component also interfaces with the hardware using a microcontroller, to execute actions such as rotating the mechanical wheels, receiving input from the players through buttons, and projecting output onto the LCD screen.

References

[1] "Blackjack: Rules of the Game." BlackjackInfo.com: Blackjack Rules, Basic Strategy, Message Boards, Card Counting, Tournament Strategy. Web. 12 Oct. 2010. <http://www.blackjackinfo.com/blackjack-rules.php>.

[2] "AUTOMATED TRACKING OF PLAYING CARDS - Google Patent Search." Google. Web. 12 Oct. 2010. <http://www.google.com/patents?id=m1mCAAAAEBAJ&printsec=abstract&zoom=4#v=onepage&q&f=false>.

[3] “Automated Poker Table - Google Patent Search." Google. Web. 12 Oct. 2010. <http://www.google.com/patents?id=64SnAAAAEBAJ&printsec=abstract&zoom=4#v=onepage&q&f=false>.”