SearchProblem

Environment kuimaze2.SearchProblem shall be used for problems of finding the shortest path in a maze. It is used in optional task 01-easy_search and the first mandatory semestral task 03-search.

After creation of the SearchProblem instance (see Usage), you can use the following methods of the environment:

• env.get_start() returns the start State of the problem.
• env.get_goals() returns the set of goal States of the problem.
• env.is_goal(state) returns True, if the state is among the goal states.
• env.get_actions(state) returns a list of all actions available in the given state.
• env.get_transition_result(state, action) returns a pair of (new_state, transition_cost) when choosing applying action in state.
• env.render() updates the graphical display of the environment. For the explanation of various parameters of this method, see help(env.render).

To use the environment, you need to:

• Import it:

  >>> from kuimaze2 import SearchProblem, Map

• Create an instance of the environment with the chosen map:

  >>> MAP = "S...G"
  >>> env = SearchProblem(Map.from_string(MAP))

• Get the start (and goals, if needed):

  >>> start = env.get_start() # start State
  >>> start
  State(r=0, c=0)
  >>> goals = env.get_goals() # set of goal States
  >>> goals
  {State(r=0, c=4)}

• Communicate with the environment:

  >>> actions = env.get_actions(start)
  >>> actions
  [<Action.UP: 0>, <Action.RIGHT: 1>, <Action.DOWN: 2>, <Action.LEFT: 3>]
  >>> next_state, cost = env.get_transition_result(start, actions[0])
  >>> next_state
  State(r=0, c=0)   # It is not possible to go UP from the start state, so stay in place
  >>> cost
  1                 # Although you stayed in place, the action costs you 1 unit
  >>> env.get_transition_result(start, actions[1])  # ... but it should be possible to go RIGHT
  (State(r=0, c=1), 1)                              # And yes, it is, we moved.

To update the graphical display of the environment (if you created it with env = SearchProblem(…, graphics=True)), use:

env.render()