ABLA - Abalone Player

Abalone Game Winning Strategies ABLA Related Links
Abalone Game
Abalone is a recent two-person strategy game. Two-player, strategy game Abalone was invented in 1990 by Laurent Levi and Michel Lalet. Abalone is introduced as "never alone" game, since the prefix "ab" means "never". Hence, it is suggested that its own name summarizes the main strategy required: "winning against loneliness". The concept of the game is based on the popular Japanese Sumo wrestling. The board is a hexagonal field representing the Sumo arena. There are 4 interlaced different size hexagons, where the most exterior hexagon has 5 circular locations at each side. In the middle there is a single location. As a total, there are 61 circular board locations. Each player has fourteen marbles that can rest in these locations. Initial game position for players is illustrated in the next Figure. Each color (black-white, or black-gray) represents a Sumo-fighter.
The board is a hexagonal field representing the Sumo arena. There are 4 interlaced different size hexagons, where the most exterior hexagon has 5 circular locations at each side. In the middle there is a single location. As a total, there are 61 circular board locations. Each player has fourteen marbles that can rest in these locations. Each color (black-white, or black-gray) represents a Sumo-fighter.

Starting from the initial configuration, each player takes a turn. During a turn, a player can shift one, two, or three marbles together in any of the six directions, provided that there is an adjacent space, i.e., in line or broadside. Furthermore, whenever a player has a numerical superiority in a line (three to two, three to one, or two to one), during a turn the player is allowed to push the opposing marbles with an inline move, even off the board. No broadside pushes are allowed.
The objective of the game is to push six opposing marbles off the edges of the board.

Rules are very simple, but numerous strategic moving, pushing and defending possibilities make the game complex. Note that, there are some variants of this game.

Winning Strategies
  • The amount of "2 or 3 in a row" marbles makes one's army's defense and offence powerful.
  • Creating a hexagon of marbles among the opponents' marbles, allows one's army to push or provide defense in all directions.
  • Moving a marble to a location where there will be no teammate neighbour is not a good strategy.
  • Forcing the opponent to keep its marbles at the edges of the board provides potential scores at any time.
  • Pushing an opponent's marble off the board might not be advantageous at all times. If such a move yields defense gaps or decreases player's attack power, it is better not to push off.
  • Make defense as tight as possible which might provide attack power in the long run, if the opponent gets careless about its defense.
  • Divide and conquer; breaking the opponent's army into two weakens the opponent's power. It is easier to deal with two weak armies, reducing the opponent's offensive ability.
Hence, the arguments mainly points out two important elements of playing Abalone well:
  • Keep the marbles around the middle of the board and force the opponent to move towards the edges.
  • Keep the marbles together as much as it is possible, to increase both offensive and defensive power.
ABLA
There are two major Abalone programs that are widely used by human players. One of them is developed by Random Software, and the other is developed by Tino Werner at the Institute for Theoretical Computer Science, in the University of Technology Graz, named as ABA-PRO. We have also implemented a simple intelligent agent for playing Abalone game, named as Abalone Player - ABLA. ABLA utilizes a minimax algorithm and alpha-beta pruning, introducing a new heuristic function. MS Visual Studio IDE is used as a programming environment with DirectX API support.

ABLA is fixed to utilize 4-ply a game tree combined with a simple heuristic. Then several games are arranged against ABA-PRO (shareware version) and Random Soft at different difficulty levels. Additionally more games are arranged by assigning the first and second turns to each program. ABLA could not be beaten even once against its opponents.

ABLA is implemented by Berk Hulagu as a graduation project in ARTI Lab.

Related Links