Maes, P. (1989). The dynamic of action selection. In Proceedings of the 1989 International Joint Conference on Artificial Intelligence, Detroit, pages 991--997.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....decision is a fundamental asset of our approach. Such automaticity is only possible because it draws on properties of the SALT model of memory [Botelho and Coelho 1995] We relate our proposal to previous work on attention shift [Beaudoin and Sloman 1993] dynamic context dependent change [Maes 1989] and commitment policies [Pollack and Ringuette 1990] and [Kinny and Georgeff 1991] and show how these former mechanisms may be defined within the proposed architecture. 1 Introduction Attention shift refers to what happens when some agent stops attending to its current thinking and attends ....

....control attention. In this latter group, some present filter overriding and commitment policies (e.g. Pollack and Ringuette 1990] Kinny and Georgeff 1991] Beaudoin and Sloman 1993] and others describe cognitive mechanisms that lend themselves to cope with the attention shift problem (e.g. [Maes 1989], Botelho and Coelho 1995] In our view, although all the policies and mechanisms presented suit certain kinds of situations, none of them is capable of handling properly the situation described in Example 1.1. In this paper we present a mechanism of attention shift based on emotion and ....

[Article contains additional citation context not shown here]

Maes, P. (1989) "The dynamics of action selection", IJCAI'89, p991-997

....areas, an attracting research field is actively emerging during last years, with a main concern: The the organization and emergence of autonomous behavior in artificial systems. During the last ten years, robotics researchers have maintained a strong debate around the action selection problem [1]. In one side, the mentalistic school has defended the need of global explicit representations of the robot, the environment and the relation between them, in order to plan ways to solve problems. This planning capability reduces to a search process on a state space [2] By searching, the robot ....

P.Maes, "The Dynamics of Action Selection", in IJCAI89, Detroit, MI, 1989, pp.991-997.

....these relationships while exploring the environment. Regarding the research on control architectures, in the last years it has been mainly focused on Behavior based architectures [3] The most representative of such architectures are Brook s subsumption architecture [5] Maes action selection [15] and Arkin s motor schema [2] Since then, many other architectures have been proposed. Liscano et al. [14] use an activity based blackboard consisting of two hierarchical layers for strategic and reactive reasoning. A blackboard database keeps track of the state of the world and a set of ....

Maes, P, 1989. The dynamics of action selection. In Proceedings of IJCAI'89, pp. 991--997.

....to a set of distributed control laws which have to be combined in order to achieve the desired task. The main challenge for these techniques is the selection of the subproblems and the formation of correct composition strategies. 1.2. 2 Behavior Based Approaches Behavior based control approaches [19, 6, 71, 7] employ a distributed control representation in order to provide reactivity in the presence of uncertainties and to overcome ubiquitous robustness and complexity limitations of monolithic control techniques. In this framework, behavior is built bottom up in a reactive fashion, driven by sensory ....

....tune their interactions every time the task requirements change. The addition of a large number of control patches , in turn, leads to behavioral proliferation and complicates the accomplishment of new tasks. Other behavior based approaches such as schema based methods [6] or action selection [71] su er from the same problems since they do not provide any tools for a principled re design if the task changes. This has led to combining basic behaviors with a sequencer to derive control as a sequences of parallel behaviors [5, 107] Here, radical changes in the task requirements can be ....

Maes, Pattie. The dynamics of action selection. In International Joint Conference on Arti cial Intelligence (Detroit, MI, 1989), IJCAII, pp. 991-997.

....FC Portugal at both EuRoboCup 2000 and RoboCup 2000. 2.1.6 Magma Freiburg Magma stands for motivation action control and goal management of agents. The action control mechanism of the Magma Freiburg team is based on extended behavior networks [25] These extend original behavior networks (see [35, 60]) to exploit information from continuous domains and to allow the concurrent execution of behaviors. Extended behavior networks consist of the following components [24, 26] Goals. These are represented by a static importance value of the goal, a goal condition describing the situation in ....

P. Maes. The Dynamics of Action Selection. In Proceedings of the Eleventh International Joint Conference on Artificial Intelligence (IJCAI-89), Detroit, 1989. Morgan Kaufmann.

....Wixson Wixson s model (to be introduced in 18.1) is basically a serial form of Nested Q learning, where we must wait for agents to terminate at their goal states before control may switch to a new agent. 15.3. 3 Maes Spreading Activation Networks Maes Spreading Activation Networks [Maes, 1989, Maes, 1989a] or Behavior Networks consist of a network of agents (or nodes) which are aware of their preconditions. Nodes can be linked to from other nodes that can help to make those preconditions come true, or be inhibited by other nodes who will cause their preconditions to not hold. They can in turn link ....

Maes, Pattie (1989), The dynamics of action selection, Proceedings of the 11th International Joint Conference on Artificial Intelligence (IJCAI-89).

.... mechanisms are possible such as those described by Norman and Long [22,23] Sloman [2,26] and Moffat and Frijda [21,20] In addition, other rep resentations for motivations and mechanisms for manipulating them have been developed at both subsymbolic and symbolic levels (e.g. by Schnepf [24] Maes [17 19] and Halperin [10] All are possible instantiations of the model described in the remainder of this paper, but the details are unimportant at present. It is enough to note that the abstract model provides the framework within which such mechanisms can be incorporated according to the particular ....

P. Maes. The dynamics of action selection. In Proceedings of the Eleventh International Joint Conference on Artificial Intelligence, pages 991--997, Detroit, 1989.

....RL algorithm we have used are explained in Section 5. The experiments are presented in Section 6. Finally, the paper concludes with Section 7. 2 RELATED WORK Since Brooks proposed the subsumption architecture [4] many other coordination mechanisms for robotic systems have been proposed (Maes [11] and Arkin [1] Regarding multi agent architectures, Liscano et al. [8] Isik [9] and Stentz [15] among others, use hierarchical centralized architectures with arbitration to decide which activity takes control of the robot. Our approach, however, is completely decentralized, which means that ....

P. Maes. The dynamics of action selection. In Proc. of IJCAI'89, pp. 991--997, 1989.

....have used are explained in Section 4. The experiments are presented in Section 5. Finally, the paper concludes with Section 6. 2. RELATED WORK Since Brooks proposed the subsumption architecture [5] many other coordination mechanisms for robotic systems have been proposed, such as those by Maes [12] and Arkin [1] Regarding multi agent architectures, Liscano et al. [9] Isik [10] and Stentz [16] among others, use hierarchical centralized architectures with arbitration to decide which activity takes control of the robot. Our approach, however, is completely decentralized, which means that ....

P. Maes. The dynamics of action selection. In Proceedings of the Eleventh International Joint Conference on Artificial Intelligence (IJCAI'89), pages 991--997, 1989.

....for the description of animals behavior; in our case, where only very primitive forms of body organization are considered, it is too restrictive. 11 are purely syntactical in the sense that the semantics only exists at the level of the intepretation by the programmer or user of the systems [Man 92, p. 100ff] Current proposals regarding mapping can be classified according to the type of structures used to encode environments. Essentially, there are symbolic approaches (whose main problem is that the handling of geometry can become very cumbersome) vector oriented approaches (the objects in the ....

....his subsumption architecture [Bro 86, 91] are adequate to construct systems required to fulfill demanding tasks in unstructured environments. For such tasks, other forms of control like meta level architectures [dP 93] or even non hierarchical, distributed methods are to be investigated [Mi 86] Ma 89] De et al. 92] 13 3.4 Motives, concepts, communication The problems human users want to have solved by autonomous systems may be tackled by a single autonomous system or in other cases more adequately by a group of cooperating autonomous systems. For both single and groups of autonomous ....

Maes, P., The Dynamics of Action Selection, in : Proc. of IJCAI-89, 991997

....original behavior networks to exploit information from continuous domains. Advantages of the original networks, such as reactivity, planning capabilities, robustness, accountance for multiple goals and its cheap and distributed calculations are maintained. 1 Extended Behavior Networks Maes [4, 5, 6] suggested a mechanism for action selection in dynamic and unpredictable domains based on so called behavior networks. Although Maes networks do work in continuous domains, they do not exploit the additional information provided by continuous states. Similarly, though there are mechanisms to ....

....goals. In [2] we have shown that the extensions proposed showed significantly higher success in the RoboCup domain [7] 1. 1 Behavior Network Description Behavior networks consist of the goals of the agent, its capabilities represented by so called competence modules and its perceptions [4]. A goal of an extended behavior network is represented by the (static) importance value of the goal, a goal condition which describes the situation in which the goal is satisfied, and a relevance condition whose truth value represents the (dynamic) relevance of the goal. 80 Relevance ....

[Article contains additional citation context not shown here]

Maes, P. (1989). The Dynamics of Action Selection. In Proceedings of the International Joint Conference on Artificial Intelligence-'89 , Morgan Kaufmann, Detroit.

....have used are explained in Section 4. The experiments are presented in Section 5. Finally, the paper concludes with Section 6. 2. RELATED WORK Since Brooks proposed the subsumption architecture [5] many other coordination mechanisms for robotic systems have been proposed, such as those by Maes [12] and Arkin [1] Regarding multi agent architectures, Liscano et al. [9] Isik [10] and Stentz [16] among others, use hierarchical centralized architectures with arbitration to decide which activity takes control of the robot. Our approach, however, is completely decentralized, which means that ....

P. Maes. The dynamics of action selection. In Proceedings of the Eleventh International Joint Conference on Artificial Intelligence (IJCAI'89), pages 991--997, 1989.

....Midfielders and offenders without the ball try to keep themselves onside and try to keep at the level of the ball within the opponents half when attacking. 7 Special Team Features Behavior selection of our agents is controlled by extended behavior networks [2, 3] that are based on work by Maes [4, 5]. Extended behavior networks consist Fig. 1. Part of a behavior network used by a soccer agent. The goals of the agent are at the top level, in the middle the competence modules and at the bottom level the situation propositions (perceptions) The complete network contains 14 competences) of the ....

Maes, P. (1989). The Dynamics of Action Selection. In Proceedings of the International Joint Conference on Artificial Intelligence-'89 , Morgan Kaufmann, Detroit.

.... mechanisms are possible such as those described by Norman and Long [22,23] Sloman [2,26] and Moffat and Frijda [21,20] In addition, other rep resentations for motivations and mechanisms for manipulating them have been developed at both subsymbolic and symbolic levels (e.g. by Schnepf [24] Maes [17 19] and Halperin [10] All are possible instantiations of the model described in the remainder of this paper, but the details are unimportant at present. It is enough to note that the abstract model provides the framework within which such mechanisms can be incorporated according to the particular ....

P. Maes. The dynamics of action selection. In Proceedings of the Eleventh International Joint Conference on Artificial Intelligence, pages 991--997, Detroit, 1989.

.... be of different type (as exemplified in the RoboCup soccer environment) ffl maintenance goals, which should be less demanding the more the goal is satisfied (e.g. have stamina ) ffl achievement goals, which should be more demanding the closer the agent is to the goal (e.g. score a goal ) Maes [ 1989; 1990; 1992 ] suggested a mechanism for action selection (MASM Maes Action Selection Mechanism [ Tyrrell, 1994 ] in dynamic and unpredictable domains based on so called behavior networks. Although MASM networks do work in continuous domains, they do not exploit the additional information ....

....next section) made it possible to reduce the number of parameters and to restrict them to the ranges printed above. This simplifies the process of finding best performing parameter values for a domain (see section 4.1) 2. 2 Activation Spreading The competence modules are connected in a network [ Maes, 1989 ] to receive activation from goals and other modules. A competence module k receives activation from a goal g i i at timestep t a t kg i 0 = fl Delta f( g i ; r t g i ) Delta ex j (1) if the module has an effect (with expectation ex j ) that is part of the goal condition and both are atoms ....

Maes, P. (1989). The Dynamics of Action Selection. In Proceedings of the International Joint Conference on Artificial Intelligence-'89 , Morgan Kaufmann, Detroit.

....charged with enacting the internal behavioral goals and 9 intentions of the agent in response to external stimuli. The output of these processes can be combined in a host of ways including, for example, priority based arbitration (subsumption) 16] behavioral fusion [4] and action selection [45] to name a few. A series of models, using schema theory, represents an agent s participation with its world. This involves the extension of our common schema theoretic approach to incorporate external, as well as internal, processes. Fortunately, schema theory is quite amenable to this strategy, ....

Maes, P., "The Dynamics of Action Selection" Proc. Eleventh Intern. Joint Conf. on Artificial Intelligence (IJCAI-89), Detroit, MI, pp. 991-997, 1989.

....collection of behaviors can perform complex tasks. Parker s work went a step further to show that cooperation can also occur between robots without explicit coordination strategies. Other researchers have evaluated certain types of coordination strategies. Maes has used spreading activation[31] [30] to arbitrate which behaviors are allowed to control the system and to interject goals into reactive systems. Behaviors are connected via activation and inhibition links with activation flowing into the system from both sensors and system goals, tending to activate agents which are both currently ....

P. Maes. The dynamics of action selection. In Proceedings Eleventh International Joint Conference on Artificial Intelligence, IJCAII-89, volume 2, pages 991-- 997, 1989.

....that can be exploited in navigation or movement control [109] The creation of temporary structures through a self organising mechanism that combines build up, break down, and feedback giving rise to autocatalysis, has been used also for other aspects of intelligent behavior. For example, Maes [64] describes an action selection system (which maybe should be better called a motivational system) 46 potential Figure 12: A potential field is a temporary structure created over an analogical representation of the world. The structure consists of vector fields which can either attract or repell ....

Maes, P. (1989) The Dynamics of Action Selection. In: Proceedings of the 11th International Joint Conference on AI (IJCAI 89) Morgan Kaufmann, Pub. Los Altos. p. 991-997.

....we need global forms of control. The research questions that we study are how adequate these hypotheses are and which activation inhibition dynamics is appropriate. To this end we are developing a series of algorithms and testing them in computer simulations. One such algorithm was discussed in (Maes, 1989). This paper describes a variation on the algorithm which is simpler and produces more interesting results 1 . Experiments have been performed for several applications. The resulting systems do exhibit the desired properties of goal orientedness, situation orientedness, adaptivity, robustness, ....

Maes, P. (1989) The Dynamics of Action Selection. Proceedings of the IJCAI-89 conference, Detroit.

No context found.

Maes, P. (1989). The dynamic of action selection. In Proceedings of the 1989 International Joint Conference on Artificial Intelligence, Detroit, pages 991--997.

No context found.

Maes, P. (1989). The dynamic of action selection. In Proceedings of the 1989 International Joint Conference on Artificial Intelligence, Detroit, pages 991-- 997.

No context found.

P. Maes, `The dynamics of action selection', Proceedings of the International Joint Conference on Artificial Intelligence, 991--997, (1989).

No context found.

P. Maes. The dynamics of action selection. In Proceedings of the Eleventh International Joint Conference on Articial Intelligence (IJCAI89) , volume 2, pages 991--998, 1989.

No context found.

Pattie Maes. The dynamics of action selection. In IJCAI-89, pages 991--997, Detroit, MI, 1989.

No context found.

Maes, P. (1989b). The dynamics of action selection. In N. Sridharan (Ed.), Proceedings of the Eleventh International Joint Conference on Artificial Intelligence (IJCAI-89): 991--997. San Mateo, CA: Morgan Kaufmann.

First 50 documents  Next 50

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC