Use of Logic in Strategy Games.

1
Use of Logic in Strategy Games.

• By Jin Yao, gtg110y
• CS 1050B

2
Close Combat 2 A Bridge too Far

• Overview
• - Close Combat 2 is a squad tactical game set in
  World War II.
• - Player is given a group of soldiers and
  vehicles to command.
• -The objective of the player is to hold key
  tactical points on the map.
• -The victor is decided by the number of key
  points held at end of game time or complete
  elimination or route of the opposing team.

3
AI elements in Close Combat 2

• Path-finding
• - The player gives each squad a command. Usually,
  these commands require the players soldier to
  move from point A to point B on the map. A
  well-coded path-finding AI will draw a path that
  is the shortest between the point and relatively
  tactically sound (in other words, not running in
  front of enemy machine gun position even though
  it was the shortest possible path).

4
Path-finding and Logic

• The path planner in CC2 uses an algorithm
  reminiscent of a paint fill in a graphics program
  that allows it to calculate the best path (as
  determined by a scoring function) from a starting
  location to every other location on the map in
  one swoop. So, for the price of determining one
  path you get all the others as well.
• Source - http//www.gameai.com/games.htmlCC2,
  interview with AI programmer of Close Combat 2.
• The basic concept of path-finding is first to
  create a group of coordinates available on the
  map.
• A algorithms, or function in logic term, is
  applied to possible coordinates.
• The algorithm uses mathematical formulas and
  functions to narrow down the possible set of
  coordinates that the group of soldiers can
  travel. These functions can be similar to ones we
  have used in our class, such as divisibility,
  Quotient-Remainder, and/or Number Theory
  (such as even/odd, prime).
• In game, the soldiers current position is taken
  in as input. These data are then used to define
  the parameter for the algorithms. Then, the
  game-engine will create every coordinates
  possible for the soldier to travel to from Point
  A to Point B. Using another algorithms, defined
  by game environment parameter (such as cover,
  threats, risks, terrain-difficulty), the possible
  path coordinates are filtered down to the best
  possible coordinate.
• This is no different from what we have done in
  class for function-relationships and number
  theory. First, a broad number of element are
  created in a set, then we apply a number of
  functions to reduce the element to desired
  elements that holds properties we want. For
  example, from a set of all real number, we may
  apply divisibility function of 4 and property of
  even number. We will end up with all real numbers
  that can be divisible by four and have property
  of even number.
• The job of AI programmer in creating path-finding
  logic is to create balanced and realistic
  functions that mimic human decision making. When
  faced with decision of going from point A to
  point B, people generally think of all possible
  way of travel. Then, based upon factors such as
  traffic condition, economic cost, and
  time-in-transit, we decide upon the best path.
  AI function in Close Combat must mimic such
  decision making through complex functions. The
  functions on the game will use factors such as
  cover from enemy fire, risks of taking fire, and
  difficulty of traversing through terrain.

5
AI elements in Close Combat 2

• Soldier Behavior
• Close Combat 2 boasts a AI that mimic real-life
  behavior and action of soldiers in World War II.
  These behaviors gives the player a more authentic
  experience. Instead of having completely
  drone-like soldiers that require constant
  commands, a well-coded AI will allow the soldiers
  to perform actions on their own and/or perform
  given order in realistic and sensible manner.

6
Conditionals and Fuzzy Logic

• I use a fuzzy logic system which weighs in
  hundreds of variables through dozens of formulas,
  eventually coming down to a probability of a
  particular action that the soldier will perform.
  To prevent inconsistant jumping of actions
  (soldier decides to go prone, then the next
  instant decides to stand up, then go prone, etc.)
  a series of weights are associated with good
  behavior. This helps reinforce a consistent set
  of good actions. In the case of bad behavior, the
  soldier is restricted from choosing a good
  behavior action until certain conditions or time
  limit has been met.
• Source - http//www.gameai.com/games.htmlCC2,
  interview with AI programmer of Close Combat 2.
• The most elementary AI behavior is conditional
  statements in the programming. In any video game,
  a task must be performed, and the game engine
  must react to the input given by the player.
• In strategy games such as Close Combat 2, these
  conditional statements are numerous. In
  battlefield, there are infinite amount of
  possible actions and consequences. The game
  designer usually limit the amount of action a
  player can make to reduce AI implementation
  requirement to reasonable level.
• In most basic step, the game engine is
  conglomerate of thousands of conditional
  statements. Many of them are simple If P, then
  Q type of conditionals.
• For example, in Close Combat 2, P , the
  supposition, can be a player command, such as
  Fire at Point X. Then the Q can be defined as
  Soldiers will shoot at point X.
• However, simple conditional statements cannot
  possibly mimic the conditions of real-life
  battlefield. To do so, AI programmer must use
  fuzzy logic.
• Using fuzzy logic, the AI programmer can have
  more room to create realistic behavior for
  soldiers. Instead of concrete true-false of
  simple conditional statement, they can create
  flexible degrees of truth for a more complex
  conditional statement.
• For example, using same situation as above, the
  supposition P has become much more detailed.
  Instead of simple player command, it has expanded
  to include situational variables, such as how far
  the target (point X) and how well the soldiers
  can shoot( aim, location). The fuzzy logic system
  then determines the possibility of soldiers
  shooting at point X. In this system, there is no
  absolute truths. Instead of having absolute
  true-false, consequences are calculated in
  probability. In the example, fuzzy logic system
  may give the game engine a certain probability
  that the soldier will fire at point X rather than
  certain answer of fire/not fire.
• So, the underlying principal of any given game is
  a conditional statement, if P then Q. However, to
  create a more interactive and sometime
  realistic game, fuzzy logic system is used to
  create the grey that mimic the real world.

7
How to beat strategy games

• So, how do we take out those Nazis in Close
  Combat 2?
• The problems with current AI.
• How to exploit those problems.

8
Fundamental problems with game AI currently.
Star Craft
Command Conquer Red Alert
9
What is the limit? How do we exploit it?

• The limit is simple. The weakness lies in the
  fact that game engines use conditional statement
  in its core principal.
• Conditional statement requires a suppositions.
  Supposition defines whether the consequence is
  true or false. Thus, in any interactive program,
  such as games, there are incredible amount of
  supposition the programmer must consider.
• As the game becomes more complex, the number of
  given supposition increase many fold. Even with
  help of fuzzy logic, there is simply too many
  decision to code in reasonable amount of time
  and resource. Therefore, the AI programmer must
  code as many decisions as possible and the game
  designer must constrict the game environment as
  much as possible to reduce the workload for AI
  programmer.
• In essence, game AI are not too different from
  canned responses to given stimulus, actions.
• A player then can exploit such situation. A
  player may spend several gaming session to get a
  feel for game AI behavior. The human player,
  will then have knowledge of AIs response to
  different situation. Knowing how the AI will
  respond, the player can then develop tactics that
  counter the AIs response. To beat the AI
  completely, rinse and repeat.

10
How to improve game AI?

• AI needs to learn.
• This is fundamental evolution in AI. A true
  intelligence must have the ability to learn. In
  logic terms, the game engine must add more
  supposition as situation arises. As the player
  counters AIs moves, AI must add more
  conditionals in its data banks. These created
  conditionals will require complex analysis of
  players behavior. However, this creates a
  fundamental problem of storage-handling of so
  many suppositions.
• AI needs creativity.
• Human are purported to have creative minds. We
  can think outside of the box. AI and logic
  cannot. Logic is defined as a system of order. It
  must follow principles and functions defined by
  mathematics and logical thinking. Creativity is
  part chaos, part logic. Chaos produce
  possibilities, while logic filters out the
  useless ones. From what we have learned in class,
  almost all of the supposition were restricted.
  Because of that restriction, logic cannot think
  outside of the box.
• All of the argument given in this presentation
  are based upon the understanding of logic in the
  course. There are perhaps many more principles
  that will defined game AI better. Perhaps,
  learning can be defined in logical term.