olufjen
commented
4 years ago Problems and corrections Pawn attack positions: WhitePawn2 Attacking c3 Color B Piece None false gamepiece Gamenone
Checking additional removals For Piece positionb2 X, Y (1, 1) wP b2PAWNName P Available positions Position: b3 W X, Y (1, 2) Position: b4 B X, Y (1, 3)
Position a3 is missing Corrected:
WhitePawn2 Attacking c3 Color B Piece None false gamepiece Gamenone
Attacking a3 Color B Piece None false gamepiece Gamenone
Checking additional removals For Piece positionb2 X, Y (1, 1) wP b2PAWNName P Available positions Position: b3 W X, Y (1, 2) Position: b4 B X, Y (1, 3)
28.05.21: The OpponentmoveProcessor must be reworked! Check the PreferredmoveProcessor. Must create actions for opponent
We have a set of chess actions that are available to choose from. We need a strategy to choose the right chess action from the set of available chess actions. To choose the best chess action we use Planning as described in chapter 10 and 11 of the AIMA book.
We need to create an initial state and a goal state (aima.core.logic.planning.State) Both the initial state and the goal state contains a set of conjunctions of ground atoms (literals). The conjunction sentence is split into its ground literals. (See the Utils class under planning) The literals may contain variables.
Examples of sentences for the init state: occupies(WhitePawn8,h2) AND OWNER(WhitePlayer,WhitePawn8)AND REACHABLE(WhitePawn8,h3) occupies(WhitePawn8,h2) AND OWNER(WhitePlayer,WhitePawn8)AND REACHABLE(WhitePawn8,h4) PROTECTEDBY(WhitePawn7,h3) AND OWNER(WhitePlayer,WhitePawn7) Examples of sentences for the goal state: occupies(WhitePawn8,h3) AND OWNER(WhitePlayer,WhitePawn8)
Each of these states is represented by a conjunction of fluents. A fluent is a Literal (aima.core.logic.fol.kb.data.Literal). A fluent is a ground functionless atom: occupies(WhitePawn8,h2) It is a relation or function that can vary from one situation (state) to the next. (p. 388). With the initial state and the goal state defined, we then create a number of ActionSchemas (aima.core.logic.planning.ActionSchema). The actionschemas represents actions. So the actions and the states become part of the Problem (aima.core.logic.planning.Problem) The schema consists of the action name, a list of all the variables used in the schema, a precondition and an effect. The precondition and effect are also represented as a conjunction of fluents. The name given to the actionschema is the same name as given to the chess action: WhitePawn6_0. which in this case means move WhitePawn6 and it is the 0 move in the sequence of moves.
All the available chess actions are possible to perform. The planning problem consists of finding which chess action to perform. All chess actions have an equivalent action schema. We need to find a strategy such that we choose the "correct" actions schema ( and chess action) to perform We have the following parameters available: The number of this move. The name of the chess pieces and the name of the chess actions. The positions of the opponent pieces. What reachable positions are protected. What reachable positions are threatened by opponent pieces. What opponent pieces can be taken.
An observation: If the HLA is given a number of action schemas and an initial and goal state that belongs to one of the action schemas. the hierarchical search function returns with the first action in the list. How to create an (HLA) Action schema so that the Problem Solver chooses the "correct" chess action? Examples of moves so far: