Proceedings Article•
Success in spades: using AI planning techniques to win the world championship of computer bridge
01 Jul 1998-pp 1079-1086
TL;DR: The latest world-championship competition for computer bridge programs was the Baron Barclay World Bridge Computer Challenge, hosted in July 1997 by the American Contract Bridge League, and the winner was a new version of Great Game Products' Bridge Baron program, which uses Hierarchical Task-Network planning techniques.
Abstract: The latest world-championship competition for computer bridge programs was the Baron Barclay World Bridge Computer Challenge, hosted in July 1997 by the American Contract Bridge League. As reported in The New York Times and The Washington Post, the competition's winner was a new version of Great Game Products' Bridge Baron program. This version, Bridge Baron 8, has since gone on the market; and during the last three months of 1997 it was purchased by more than 1000 customers.The Bridge Baron's success also represents a significant success for research on AI planning systems, because Bridge Baron 8 uses Hierarchical Task-Network (HTN) planning techniques to plan its declarer play. This paper gives an overview of those techniques and how they are used.
Citations
More filters
TL;DR: The SHOP2 planning system as discussed by the authors received one of the awards for distinguished performance in the 2002 International Planning Competition and described the features that enabled it to excel in the competition, especially those aspects of SHOP 2 that deal with temporal and metric planning domains.
Abstract: The SHOP2 planning system received one of the awards for distinguished performance in the 2002 International Planning Competition. This paper describes the features of SHOP2 which enabled it to excel in the competition, especially those aspects of SHOP2 that deal with temporal and metric planning domains.
838 citations
[...]
TL;DR: The past successes, current projects, and future research directions for AI using computer games as a research test bed are reviewed.
Abstract: In 1950, Claude Shannon published his seminal work on how to program a computer to play chess. Since then, developing game-playing programs that can compete with (and even exceed) the abilities of the human world champions has been a long-sought-after goal of the AI research community. In Shannon's time, it would have seemed unlikely that only a scant 50 years would be needed to develop programs that play world-class backgammon, checkers, chess, Othello, and Scrabble. These remarkable achievements are the result of a better understanding of the problems being solved, major algorithmic insights, and tremendous advances in hardware technology. Computer games research is one of the important success stories of AI. This article reviews the past successes, current projects, and future research directions for AI using computer games as a research test bed.
96 citations
Cites background or methods from "Success in spades: using AI plannin..."
...In the 1990s, several academic efforts began using bridge for research in AI (Frank 1998; Ginsberg 1999; Smith, Nau, and Throop 1998a, 1998b; Ginsberg 1996b)....
[...]
...Rather than build a search tree where each branch was the play of a card, they would define each branch as a strategy, using human-defined concepts such as finesse and squeeze (Smith, Nau, and Throop 1998a, 1998b)....
[...]
Proceedings Article•
01 Jul 1998TL;DR: This paper describes how to overcome the difficulties that can result from the use of backward chaining and partial-order planning by adapting Hierarchical Task-Network planning to use a total-order control strategy that generates the steps of a plan in the same order that those steps will be executed.
Abstract: AI planning techniques are beginning to find use in a number of practical planning domains. However, the backward-chaining and partial-order-planning control strategies traditionally used in AI planning systems are not necessarily the best ones to use for practical planning problems. In this paper, we discuss some of the difficulties that can result from the use of backward chaining and partial-order planning, and we describe how these difficulties can be overcome by adapting Hierarchical Task-Network (HTN) planning to use a total-order control strategy that generates the steps of a plan in the same order that those steps will be executed. We also examine how introducing the total-order restriction into HTN planning affects its expressive power, and propose a way to relax the total-order restriction to increase its expressive power and range of applicability.
92 citations
Cites background or methods from "Success in spades: using AI plannin..."
...This decomposition hierarchy can be represented quite naturally using HTNs, making it easy to develop plan details depending on the details of the design....
[...]
...1994; Hebbar et al. 1996; Smith et al. 1997; Smith et al. 1998; Wilkins & Desimone 1994)....
[...]
...More information about Bridge Baron 8 appears elsewhere in this conference proceedings (Smith et al 1998)....
[...]
Proceedings Article•
01 Jun 2005TL;DR: HTN planners can be used to generate correct plans for coordinated team AI behavior modeled with TMK representations, and TMK models are of interest to game AI because, as it is shown, they are as expressive as HTNs but have more convenient syntax.
Abstract: In this paper we explore the use of Hierarchical-Task-Network (HTN) representations to model strategic game AI. We will present two case studies. The first one reports on an experiment using HTNs to model strategies for Unreal Tournament® (UT) bots. We will argue that it is possible to encode strategies that coordinate teams of bots in first-person shooter games using HTNs. The second one compares an alternative to HTNs called Task-Method-Knowledge (TMK) process models. TMK models are of interest to game AI because, as we will show, they are as expressive as HTNs but have more convenient syntax. Therefore, HTN planners can be used to generate correct plans for coordinated team AI behavior modeled with TMK representations.
86 citations
Cites background from "Success in spades: using AI plannin..."
...For example, Bridge Baron® 8 won the 1997 world-championship competition for computer programs using HTN planning techniques to plan its declarer play (Smith et al., 1998)....
[...]
30 Jul 2000
TL;DR: The past, present, and future of the development of games-playing programs are discussed and some surprising changes of direction occurring that will result in games being more of an experimental testbed for mainstream AI research, with less emphasis on building world-championship-caliber programs.
Abstract: The development of high-performance game-playing programs has been one of the major successes of artificial intelligence research. The results have been outstanding but, with one notable exception (Deep Blue), they have not been widely disseminated. This talk will discuss the past, present, and future of the development of games-playing programs. Case studies for backgammon, bridge, checkers, chess, go, hex, Othello, poker, and Scrabble will be used. The research emphasis of the past has been on high performance (synonymous with brute-force search) for twoplayer perfect-information games. The research emphasis of the present encompasses multi-player imperfect/nondeterministic information games. And what of the future? There are some surprising changes of direction occurring that will result in games being more of an experimental testbed for mainstream AI research, with less emphasis on building world-championship-caliber programs. One of the most profound contributions to mankind’s knowledge has been made by the artificial intelligence (AI) research community: the realization that intelligence is not uniquely human. 1 Using computers, it is possible to achieve human-like behavior in nonhumans. In other words, the illusion of human intelligence can be created in a computer. This idea has been vividly illustrated throughout the history of computer games research. Unlike most of the early work in AI, game researchers were interested in developing high-performance, real-time solutions to challenging problems. This led to an ends-justify-the-means attitude: the result—a strong chess program—was all that mattered, not the means by which it was achieved. In contrast, much of the mainstream AI work used simplified domains, while eschewing real-time performance objectives. This research typically used human intelligence as a model: one only had to emulate the human example to achieve intelligent behavior. The battle (and philosophical) lines were drawn. The difference in philosophy can be easily illustrated. The human brain and the computer are different machines, each with its own sets of strengths and weaknesses. Humans are good at, for example, learning, reasoning by analogy, and
60 citations
References
More filters
Proceedings Article•
01 Aug 1996TL;DR: In this paper, a combination of a generative planning tool and an influence diagram solver is used to solve a real world planning problem, where a planning agent manages constraints that order sets of feasible equipment employment actions, and an optimum solution to the employment problem based on the objective function is found.
Abstract: This paper works through the optimization of a real world planning problem, with a combination of a generative planning tool and an influence diagram solver The problem is taken from an existing application in the domain of oil spill emergency response The planning agent manages constraints that order sets of feasible equipment employment actions This is mapped at an intermediate level of abstraction onto an influence diagram In addition, the planner can apply a surveillance operator that determines observability of the state--the unknown trajectory of the oil The uncertain world state and the objective function properties are part of the influence diagram structure, but not represented in the planning agent domain By exploiting this structure under the constraints generated by the planning agent, the influence diagram solution complexity simplifies considerably, and an optimum solution to the employment problem based on the objective function is found Finding this optimum is equivalent to the simultaneous evaluation of a range of plans This result is an example of bounded optimality, within the limitations of this hybrid generative planner and influence diagram architecture
8 citations
01 Jan 1997
TL;DR: In this article, a new variant of Hierarchical Task Network (HTN) planning, called planning using TOFS (Total-Order Forward Search), has been developed, which is used to solve declarer's cards at contract bridge.
Abstract: Because most real-world planning problems are difficult, AI planning researchers have needed to make simplifying assumptions in order to solve some of these problems at all. These simplifying assumptions eliminated some of the difficult features that need to be considered to solve other problems. For example, in existing AI planning algorithms, the approaches to dealing with uncertainty and numerical values are insufficient to handle many important problems.
To address these limitations, I have developed a new variant of Hierarchical Task Network (HTN) planning, which I call planning using TOFS (Total-Order Forward Search). In TOFS, a planner always instantiates the operators in a plan in the order that they will be executed.
I have applied TOFS successfully to two very different real-world problems. (1) The first problem was play of declarer's cards at contract bridge. My implementation of HTN planning using TOFS, called Tignum 2, does statistically significantly better than the best available computer bridge program. (2) The second problem was automated process planning for the manufacture of microwave modules. My implementation of HTN planning using TOFS, called the EDAPS Process Planner, incorporates electronic and mechanical manufacturing processes, works concurrently with an electronic CAD tool, and provides feedback about manufacturability and lead time to the designers, based on actual process plans for the manufacture of the device. In both domains, HTN planning using TOFS ran relatively quickly, and searched relatively few nodes.
In this dissertation, I describe HTN planning using TOFS, its application to these two domains, and consequent observations. I detail the adaptations of HTN planning using TOFS required to apply it to bridge, and the adaptations required to apply it to microwave module manufacture. I observe that TOFS allows the preconditions of methods to be written with arbitrary computer code, and discuss the advantages and disadvantages of this technique.
Because both research projects were successful, I am confident that HTN planning using TOFS is an important new AI planning methodology. To reap the benefits of HTN planning using TOFS, a classification of domains into those for which it is suitable and those for which it is unsuitable must be further developed.
5 citations
"Success in spades: using AI plannin..." refers background or methods or result in this paper
...Furthermore, the same adaptation of HTN planning that we used for computer bridge is also proving useful for the generation and evaluation of manufacturing plans for microwave transmit/r eceive modules, as part of a project that some of us have with Northrop Grumman Corporation (Hebbar et al. 1996; Smith et al. 1996b; Smith et al. 1996d; Smith 1997)....
[...]
...The size of this tree would vary depending on the particular bridge deal—but it would include about 5.6x1044 leaf nodes in the worst case (Smith 1997, p. 226), and about 2.3x1024 leaf nodes in the average case (Lopatin 1992, p. 8)....
[...]
...Our approach (Smith et al. 1996a; Smith et al. 1996c; Smith et al. 1996e; Smith 1997) grew out of the observation that bridge is a game of planning....
[...]
...…we used for computer bridge is also proving useful for the generation and evaluation of manufacturing plans for microwave transmit/receive modules, as part of a project that some of us have with Northrop Grumman Corporation (Hebbar et al. 1996; Smith et al. 1996b; Smith et al. 1996d; Smith 1997)....
[...]
...Furthermore, the same adaptation of HTN planning that we used for computer bridge is also proving useful for the generation and evaluation of manufacturing plans for microwave transmit/receive modules, as part of a project that some of us have with Northrop Grumman Corporation (Hebbar et al. 1996; Smith et al. 1996b; Smith et al. 1996d; Smith 1997)....
[...]
01 Jan 1996
TL;DR: In this article, the authors present a planning system for an important industrial planning problem, which is based on a modified version of HTN planning, and provide an overview of its operation, and compare and contrast it to how HTN plans are normally done.
Abstract: This paper reports on the development of a planning system for an important industrial planning problem. This planner, which is one of the modules in an integrated design-and-planning system called EDAPS, provides an integrated approach to process planning in both the electronic and mechanical domains, specifically in the manufaeture of microwave transmit-receive O/R) modules. Our planner is based on a modified version of HTN planning. We provide an overview of its operation, and compare and contrast it to how HTN planning is normally done. This same modified HTN planning strategy appears to be useful in a variety of application domains. For example, as described in (Smith, Nau, & Throop 1996), the basic approach has been used successfully for declarer play in the game of contract bridge. We discuss why this approach is successful in two such diverse application domains.
3 citations