Content deleted Content added
fixed typo: synthesise -> synthesize |
m →See also: alphabetical order |
||
| (36 intermediate revisions by 24 users not shown) | |||
Line 1:
{{Short description|Branch of artificial intelligence}}
{{More footnotes|date=January 2012}}
{{Artificial intelligence|Major goals}}
'''Automated planning and scheduling''', sometimes denoted as simply '''AI planning''',<ref>{{Citation | last1=Ghallab |
In known environments with available models, planning can be done offline. Solutions can be found and evaluated prior to execution. In dynamically unknown environments, the [[strategy]] often needs to be revised online. Models and policies must be adapted. Solutions usually resort to iterative [[trial and error]] processes commonly seen in [[artificial intelligence]]. These include [[dynamic programming]], [[reinforcement learning]] and [[combinatorial optimization]]. Languages used to describe planning and scheduling are often called [[action language]]s.
== Overview ==
{{see|State space search}}
{{More citations needed|section
| name = More citations needed section
| find = none
|date=February 2021
}}
Given a description of the possible initial states of the world, a description of the desired goals, and a description of a set of possible actions, the planning problem is to synthesize a plan that is guaranteed (when applied to any of the initial states) to generate a state which contains the desired goals (such a state is called a goal state).
The difficulty of planning is dependent on the simplifying assumptions employed. Several classes of planning problems can be identified depending on the properties the problems have in several dimensions.
* Are the actions [[deterministic]] or
* Are the [[state
* Can the current state be observed unambiguously? There can be full observability and partial observability.
* How many initial states are there, finite or arbitrarily many?
* Do actions have a duration?
* Can several actions be taken concurrently, or is only one action possible at a time?
* Is the objective of a plan to reach a designated goal state, or to maximize a [[reward function]]?
* Is there only one agent or are there several agents? Are the agents cooperative or selfish? Do all of the agents construct their own plans separately, or are the plans constructed centrally for all agents?
Line 38 ⟶ 46:
* and a single agent.
When full observability is replaced by partial observability, planning corresponds to a [[partially observable Markov decision process]] (POMDP).
If there are more than one agent, we have [[multi-agent planning]], which is closely related to [[game theory]].
Line 44 ⟶ 52:
== Domain independent planning ==
{{More citations needed|section
In AI planning, planners typically input a ___domain model (a description of a set of possible actions which model the ___domain) as well as the specific problem to be solved specified by the initial state and goal, in contrast to those in which there is no input ___domain specified. Such planners are called "___domain independent" to emphasis the fact that they can solve planning problems from a wide range of domains. Typical examples of domains are block stacking, logistics, workflow management, and robot task planning. Hence a single ___domain independent planner can be used to solve planning problems in all these various domains. On the other hand, a route planner is typical of a ___domain specific planner.▼
| name = More citations needed section
| find = none
|date=February 2021
}}
▲In AI planning, planners typically input a ___domain model (a description of a set of possible actions which model the ___domain) as well as the specific problem to be solved specified by the initial state and goal, in contrast to those in which there is no input ___domain specified. Such planners are called "___domain independent" to
== Planning ___domain modelling languages{{anchor|Planning_languages}} ==
{{More citations needed|section
| name = More citations needed section
| find = none
|date=February 2021
}}
The most commonly used languages for representing planning domains and specific planning problems, such as [[Stanford Research Institute Problem Solver|STRIPS]] and [[Planning Domain Definition Language|PDDL]] for Classical Planning, are based on state variables. Each possible state of the world is an assignment of values to the state variables, and actions determine how the values of the state variables change when that action is taken. Since a set of state variables induce a state space that has a size that is exponential in the set, planning, similarly to many other computational problems, suffers from the [[curse of dimensionality]] and the [[combinatorial explosion]].
Line 60 ⟶ 80:
{{See also|Sussman anomaly}}
===Action model learning===
Creating ___domain models is difficult, takes a lot of time, and can easily lead to mistakes. To help with this, several methods have been developed to automatically learn full or partial ___domain models from given observations.
<ref>{{cite conference |author=Callanan, Ethan and De Venezia, Rebecca and Armstrong, Victoria and Paredes, Alison and Chakraborti, Tathagata and Muise, Christian |title=MACQ: A Holistic View of Model Acquisition Techniques |conference=ICAPS Workshop on Knowledge Engineering for Planning and Scheduling (KEPS) |year=2022 |url=https://icaps22.icaps-conference.org/workshops/KEPS/KEPS-22_paper_4962.pdf}}</ref>
<ref>{{cite journal |author=Aineto, Diego and Jiménez Celorrio, Sergio and Onaindia, Eva |title=Learning action models with minimal observability |journal=Artificial Intelligence |volume=275 |pages=104–137 |year=2019 |doi=10.1016/j.artint.2019.05.003 |url=https://doi.org/10.1016/j.artint.2019.05.003|doi-access=free |hdl=10251/144560 |hdl-access=free }}</ref>
<ref>{{cite journal |author=Jiménez, Sergio and de la Rosa, Tomás and Fernández, Susana and Fernández, Fernando and Borrajo, Daniel |title=A review of machine learning for automated planning |journal=The Knowledge Engineering Review |volume=27 |issue=4 |pages=433–467 |year=2012 |doi=10.1017/S026988891200001X |url=https://doi.org/10.1017/S026988891200001X|url-access=subscription }}</ref>
*Read more: [[Action model learning]]
=== Reduction to other problems ===
* reduction to the [[propositional satisfiability]] problem ([[satplan]]).
* reduction to [[
===Temporal planning===
Temporal planning can be solved with methods similar to classical planning. The main difference is, because of the possibility of several, temporally overlapping actions with a duration being taken concurrently, that the definition of a state has to include information about the current absolute time and how far the execution of each active action has proceeded. Further, in planning with rational or real time, the state space may be infinite, unlike in classical planning or planning with integer time. Temporal planning is closely related to [[scheduling]] problems when uncertainty is involved and can also be understood in terms of [[timed automaton|timed automata]]. The Simple Temporal Network with Uncertainty (STNU) is a scheduling problem which involves controllable actions, uncertain events and temporal constraints. Dynamic Controllability for such problems is a type of scheduling which requires a temporal planning strategy to activate controllable actions reactively as uncertain events are observed so that all constraints are guaranteed to be satisfied. <ref>{{cite journal |last1=Vidal |first1=Thierry |title=Handling contingency in temporal constraint networks: from consistency to controllabilities |journal=Journal of Experimental & Theoretical Artificial Intelligence |date=January 1999 |volume=11 |issue=1 |page=23--45 |doi=10.1080/095281399146607 |citeseerx=10.1.1.107.1065 }}</ref>
===Probabilistic planning===
Line 81 ⟶ 108:
===Conditional planning===
Deterministic planning was introduced with the [[Stanford Research Institute Problem Solver|STRIPS]] planning system, which is a hierarchical planner. Action names are ordered in a sequence and this is a plan for the robot. Hierarchical planning can be compared with an automatic generated [[Behavior tree (artificial intelligence, robotics and control)|behavior tree]].<ref>{{cite journal |title=Building a Planner: A Survey of Planning Systems Used in Commercial Video Games |author=Neufeld, Xenija and Mostaghim, Sanaz and Sancho-Pradel, Dario and Brand, Sandy |journal=IEEE Transactions on Games |year=2017 |publisher=IEEE}}</ref> The disadvantage is, that a normal behavior tree is not so expressive like a computer program. That means, the notation of a behavior graph contains action commands, but no [[Loop (computing)|loops]] or if-then-statements. Conditional planning overcomes the bottleneck and introduces an elaborated notation which is similar to a [[control flow]], known from other programming languages like [[Pascal (programming language)|Pascal]]. It is very similar to [[program synthesis]],
An early example of a conditional planner is “Warplan-C” which was introduced in the mid 1970s.<ref>{{cite conference |title=Conditional nonlinear planning |author=Peot, Mark A and Smith, David E |conference=Artificial Intelligence Planning Systems |pages=189–197 |year=1992 |publisher=Elsevier|url=https://sites.google.com/site/markpeot2/peot92conditional.pdf}}</ref> What is the difference between a normal sequence and a complicated plan, which contains if-then-statements? It has to do with uncertainty at [[Run time (program lifecycle phase)|runtime]] of a plan. The idea is that a plan can react to [[Soft sensor|sensor signals]] which are unknown for the planner. The planner generates two choices in advance. For example, if an object was detected, then action A is executed, if an object is missing, then action B is executed.<ref>{{cite conference |title=Conditional progressive planning under uncertainty |author=Karlsson, Lars |conference=IJCAI |pages=431–438 |year=2001|url=https://www.researchgate.net/publication/2927504}}</ref> A major advantage of conditional planning is the ability to handle [[Partial-order planning|partial plans]].<ref>{{cite
====
We speak of "contingent planning" when the environment is observable through sensors, which can be faulty. It is thus a situation where the planning agent acts under incomplete information. For a contingent planning problem, a plan is no longer a sequence of actions but a [[decision tree]] because each step of the plan is represented by a set of states rather than a single perfectly observable state, as in the case of classical planning.<ref>{{Cite conference|conference=International Joint Conference of Artificial Intelligence (IJCAI)|year=2009|author1=
==== Conformant planning ====
Conformant planning is when the agent is uncertain about the state of the system, and it cannot make any observations. The agent then has beliefs about the real world, but cannot verify them with sensing actions, for instance. These problems are solved by techniques similar to those of classical planning,<ref>{{Cite journal|title=Compiling uncertainty away in conformant planning problems with bounded width|journal=Journal of Artificial Intelligence Research|
== Deployment of planning systems ==
* The [[Hubble Space Telescope]] uses a short-term system called [https://archive.
== See also ==
* [[Action description language]]
* [[Action model learning]]
* [[Actor model]]
* [[Applications of artificial intelligence]]
* [[Constraint satisfaction problem]]
* [[International Conference on Automated Planning and Scheduling]]
* [[Reactive planning]]
* [[Scheduling (computing)]]
Line 107 ⟶ 136:
; Lists
* [[List of SMT solvers]]▼
* [[List of constraint programming languages]]
* [[List of emerging technologies]]
▲* [[List of SMT solvers]]
* [[Outline of artificial intelligence]]
Line 116 ⟶ 146:
==Further reading==
* {{cite journal|url=http://www.eetn.gr/index.php/eetn-publications/ai-research-in-greece/planning-and-scheduling |last=Vlahavas |first=I |title=Planning and Scheduling |journal=EETN |url-status=dead |
== External links ==
| |||