-
Notifications
You must be signed in to change notification settings - Fork 0
Historical Timeline (Software)
Many reviews of AI planning technology have been written. Some are listed here.
AI Planning: Systems and Techniques, James Hendler, Austin Tate, and Mark Drummond, AI Magazine Volume 11 Number 2 (1990) - PDF Document.
General Problem Solver (GPS) is a computer program created in 1959 by Herbert A. Simon, J. C. Shaw, and Allen Newell (RAND Corporation) intended to work as a universal problem solver machine.
The Stanford Research Institute Problem Solver, known by its acronym STRIPS, is an automated planner developed by Richard Fikes and Nils Nilsson in 1971 at SRI International.
Traverser (also called Graph Traverser 4 - GT4) was created by Austin Tate during 1971-1972 in an undergraduate student project in Computer Studies at the University of Lancaster. It built on the work of Donald Michie and his colleagues at the Department of Machine Intelligence and Perception at the University of Edinburgh. Written in POP-2.
A planner which uses a library of methods to create a plan and can debug failures to suggest fixes. Created by Gerry Sussman at the MIT AI Lab.
STRIPS with abstraction layers for activity representation. Earl Sacerdoti, SRI International and Stanford University.
An early planning system whose search was directed by finding and debugging "approaches" that handled the goal structure underlying the problem. Interplan was created in 1972-1975 as part of Austin Tate's PhD, supervised by Professor Donald Michie at the Machine Intelligence Research Unit at the University of Edinburgh. Written in POP-2.
Nets of Action Hierarchies. An early AI planner using hierarchical activity representation. Used as part if an engineering task support assistant. Earl Sacerdoti, SRI International.
An early hierarchical task network, partial-order planner - used as the basis for text book descriptions of this type of technology (Russell & Norvig). Nonlin was created in 1975-1976 at Department of Artificial Intelligence at the University of Edinburgh during the first year of the UK Science Research Council project entitled "Planning: a joint AI/OR Approach" whose Principal Investigator was Professor Bernard Meltzer. Austin Tate was the AI researcher on the project and Lesley Daniel was the Operational Research researcher. Nonlin continued to be developed in the 1982-3 period in discussions with Steve Vere at the NASA Jet Propulsion Laboratory (who created the DEVISER planner), and in joint work with Andrew Whiter at Systems Designers Ltd in the UK. Brian Drabble's Excalibur planner was built on Nonlin by adding a qualitative process reasoner. Writen in POP-2.
"The 'Deviser III' automated spacecraft mission planner prototype can plan complete Voyager spacecraft operational sequences consisting of over 100 data capture goals, working 10-50 times faster than a human analyst. The planner generates a loop-free network of actions and events resembling a program evaluation and review technique chart. Actions, events, and inferences are the building blocks from which plans are assembled. It is noted that, for a very large spacecraft mission plan, the backtracking through remembered decision trees entailed by this artificial intelligence system can consume megabyte quantities of computer memory."
An AI planning, execution, and plan repair system with extensive representations for temporal, resource and other constraints, support for plan execution monitoring and plan repair "on-the-fly". It was created at the Artificial Intelligence Applications Institute (AIAI) at the University of Edinburgh. O-Plan's representation uses the <I-N-OVA> model. Runnable as a Web Service. Written in Common Lisp with User Interface and Web Service elements in Java.
SIPE (System for Interactive Planning and Execution Monitoring) planning system created by David Wilkins at SRI International.
Brian Drabble's Ph.D project Excalibur planner linked HTN planning methods with qualitative process reasoning. It was based partly on [Nonlin].
Graphplan is an algorithm for automated planning developed by Avrim Blum and Merrick Furst in 1995. Graphplan takes as input a planning problem expressed in STRIPS and produces, if one is possible, a sequence of operations for reaching a goal state.
Simple Hierarchical Ordered Planner. Built by Dana Nau and his team at the University of Maryland.
A portable cross-platform Java-based planning and collaboration support environment. It was created at the Artificial Intelligence Applications Institute (AIAI) at the University of Edinburgh. Aid to multi-agent cooperative work and external services use. Plan representation based on the <I-N-C-A> conceptual model. Written in Java.
The Lightweight Automated Planning Toolkit (LAPKT) separates search engines from the data structures used to represent planning tasks. This second component receives the name of 'interface' since it is indeed the interface that provides the search model to be solved.
Tarski is a framework for the specification, modeling and manipulation of AI planning problems. Tarski is written in Python and includes parsers for major modeling languages (e.g., PDDL, FSTRIPS, RDDL), along with modules to perform other common tasks such as logical transformations, reachability analysis, grounding of first-order representations and problem reformulations.