Examples Of Automated Planning

1162 Words3 Pages

section{Planning and Scheduling} label{sec:planingscheduling}

Automated planning is concerned with making a plan for solving a problem. When working with these kind of problems, the difficulty primarily lies in defining the problem in a precise yet relatively simple way.

There are different approaches on how to do this, based on what kind of problem needs to be solved. Often their representation will include definitions of states, actions and functions which map states to a new state. The state describes the current world situation. When an action occurs the situation is affected and transits into other states. States, actions and functions are described as mathematical setscite{AutomatedPlanning}:
% is a scheduling system - Jalil
egin{itemize}
item $Sigma = (S, A, gamma)$ item $S: states$ item $A: actions$ item $gamma$: state transition function $(gamma: S imes A ightarrow S)$ end{itemize} A planning problem is often defined as:
egin{equation}
P = (Sigma, s_0, g) end{equation} Where $s_0$ describes the initial state and $g$ describes the desired goal state. The idea is to find a transition from $s_0$ to $g$ through different actions. If all the states have to be explicitly defined, it will quickly grow to an extreme amount of definitions of states. To solve this problem, new states are defined dynamically through the actions applied to the states.

subsection{Set-Theoretic Representation}label{subsec:settheoretic}
The set-theoretic representation is one of the ways to represent a planning problem. Given a finite set $L$ of propositions, we can describe the environment as following:
egin{equation}
L = {p_1, ..., p_n} end{equation} An example of a proposition could be a function named f...

... middle of paper ...

...ith $gamma^-1(s, a)$. It is the same principle as finding the the successors of a state, but instead it goes the opposite direction and finds the states leading to a state.
%egin{itemize}
% item $gamma^{-1}(g, a) (g - effects^+(a)) cup precond(a)$
% item $gamma(s, a) in s_g $ iff $ gamma^{-1}(g, a) subseteq s$
% item $Gamma^{-1}(g) = {Gamma^{-1}(g, a) mid a in A $ is relevant for $ g}$
% item Supersets reachable of g: $hat{Gamma}^{-1}(g) = Gamma^{-1}(g) cup Gamma^{-2}(g) cup….$
%end{itemize}

%egin{itemize}
% item[] P has solution iff $S_g cap hat{Gamma} eq emptyset$
% item[] P has solution iff s0 is a superset of some element in $hat{Gamma}^{-1}(g)$
% item[] P and P' have the same statement then both P and P' have the same set of reachable states $hat{Gamma}(s0)$ and the same set of solutions.
%end{itemize}

Open Document