problem types(slides 3) Flashcards
single-state
deterministic, fully observable
-Agent knows exactly which state it will be in; solution is a sequence
sensorless problem(conformant problem)
Non-observable
-Agent may have no idea where it is; solution is a sequence
contingency problem
Nondeterministic and/or partially observable
- percepts provide new information about current state
- often interleave search, execution
exploration problem
unknown state space
search
the process of looking for a fixed sequence of actions
search algorithm
input: problem
output: an action sequence
the four items of a problem
initial state, actions, transition model, goal test
successor(s)
states reachable from s by performing an actioin
transition model
Result(s,a) returns the state resulting by performing action a in state a
goal test
allows to check if state is a goal
- explicit: eg “at Bucharest”
- implicit: eg “Checkmate(x)”
Abstraction
state space must be abstracted from real world for problem solving
Abstract state= set of real states
same with abstract action
abstract solution= set of real paths that are solutions in the real world