D. W. Payton, J. K. Rosenblatt, and D. M. Keirsey. Plan guided reaction. IEEE Trans. on Systems, Man, and Cybernetics, 20(6):1370--1382, 1990.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... been developed in an attempt to address the shortcomings of purely reactive robotic architectures [1,5,12] Other approaches attempt to morph higherlevel plans into forms suitable for reactive behaviors [8, 9, 10] Payton also delineates a method for using information in a reactive architecture [14]. In this method, a priori knowledge becomes an enabling resource for decisionmaking. From his perspective, traditional plans are artificially abstracted from knowledge that often results in over or under specification of a mission s objectives. By minimizing symbolic abstraction, one can develop ....

....robot controller that contains the collection of other active behaviors, such as move to goal, avoidobstacles, noise, etc. This tends to smooth the digitization bias present in the internalized plan. Figure 2 depicts a typical internalized plan. 2. 3 Plans in Parallel As discussed by Payton [14], internalized plans offer an advantage in that several plans can be utilized simultaneously. We incorporate multiple plans in two ways: in parallel and in serial. Figure 3 depicts the use of parallel internalized plans. Serial plans are discussed in Section 2.4. Parallel plans consist of more ....

D. Payton, J. Rosenblatt, and D. Keirsey, "Plan Guided Reaction" In IEEE Transactions on Systems Man and Cybernetics, pp. 1370-1382, 20(6), 1990.

....that is, to extract from the original image some features computed on a number of images regions. 1. Introduction Over the recent years, neural networks based algorithms for robot motion control by digital image processing have been developed. A list of relevant recent references is given by [6, 7, 8, 9, 13, 14, 15, 19, 20]. It is evident that giving as input directly the image would mean to associate an input neuron to each image pixel, leading to very complex networks also in the case of low dimensional images. Moreover, it is intuitive to gather that within a given image, the different portions carry a different ....

D. PAYTON, J. ROSEMBLATT, AND D. KEIRSEY, Plan Guided Reaction, IEEE Trans. on Systems, Man, and Cybernetics, Vol. 20, No. 6, pp. 1370 -- 1382, Oct. 1993

....the planner is invoked and guided by conditions in the sequencing layer, e.g. a task failing or completing. There are many different instantiations of this architecture, including Sapphira [Kon97] SSS [Con92] ATLANTIS [Gat92] RAPs [Fir94] AuRA [Ark90] and Payton s reactive planners [Pay90]. In almost all of these, the sequencer plays the role of the main executive, taking advice from the planner and invoking behaviors to accomplish goals. When one thinks of writing robot programs, it is sequencer programs that are the result. In fact, it s possible to think of the planner as an ....

D. W. Payton, J. K. Rosenblatt, and D.M. Keirsey, Plan guided reaction, IEEE Trans. on Systems, Man, and Cybernetics 20 (6), 1990. Colbert Executive 01/22/99 1

....the knowledge of the actions they will perform to define a set of simple vision modules. In practice both of these systems tend to be coarse grained, macromodule systems. By the late 1980 s a number of researchers began presenting their ideas for new architectures [2] 3] 9] 10] 16] and [31]. These principle problem for these designs to address was that planning, perception and action are computationally expensive operations. To achieve real time performance it was clear that processing had to be distributed or somehow simplified. Some researchers made the observation that more ....

....module was to accept messages from the other modules, to decide how to attend to a variety of goals and to decide what to do if things were not going according to plan. It would, for example, inform the planner that its plans are failing or no longer applicable. Payton, Rosenblatt and Keirsey [31] observed that fine grained interaction was key. They designed a system consisted of a collection of fine grained behaviors, designed using connectionist ideas, guided by what they called internalized plans. Internalized plans, as they characterize them, are like gradient fields that, at each ....

Panton, D., Rosenblatt, J. K., and Keirsey, D. M. "Plan Guided Reaction." IEEE Transactions on Systems, Man and Cybernetics, Vol. 20, No. 6, pp. 1370-1382, November/December 1990.

....physical objects whereas behavior based systems usually operate on raw sensory information with limited processing. How to close the gap between planning and behavior based system components is thus an important issue, which has received a considerable amount of attention in the literature, e.g. (Payton, Rosenblatt Keirsey 1990, Arkin 1989, Gat 1992, Firby 1992, Saffiotti 1993, Pirjanian, Blaasvaer Christensen 1994, Bonnaso, Gat, Firby Kortenkamp 1996) Some of the work in hybrid systems deals with integrating planning and reactive behavior based system components by introducing, e.g. mediators that close the gap ....

.... et al. 1997) Examples of cooperative mechanisms are the potential field method (Khatib 1986) motor schemas used in (Arkin 1989) blending of fuzzy behaviors (Saffiotti et al. 1995, Yen Pfluger 1995) the dynamic systems approach (Large, Christensen Bajcsy 1997) and voting mechanisms (Payton et al. 1990, Rosenblatt 1995) A similar approach to voting is introduced in (Riekki Roning 1997) by combining action maps. 1.3 Related work overview 9 In the potential field approach (Khatib 1986) outputs generated by behaviors are interpreted as potential fields (attractive and repulsive forces) and ....

[Article contains additional citation context not shown here]

Payton, D. W., Rosenblatt, J. K. & Keirsey, D. M. (1990), `Plan guided reaction', IEEE transactions on Systems, Man, and Cybernetics 20(6), 1370--1382.

....paper described below. Khatib, Khatib,1985] develops a method for real time obstacle avoidance for manipulators and variations of this method have been used in a number of papers. Some of them that are also important from the perspective of reactivity have been included in this study such as [Payton,1990, Arkin,1988, Arkin,1987] Khatib s paper is important to this study because it defines a method for modeling the world and this can be developed, as shown in this paper, into a reactive control strategy.Since, this is a method to model the world, it can be applied as an algorithm that construct ....

....there is a requirement for a plan, that is based on a global observation of the world. This point was mentioned in out discussion on Brook s and Payton s work. At the same time the pitfalls of classical planning based on a world model 37 (which cannot be complete for a real world) are known. [Payton,1990] [Kaebling,89] explain this concept in detail and propose a plan guided reaction as a paradigm for intelligent systems. ffl Generating robust models of the world incrementally The primary problem with using potential fields to model the world is that the potential field may contain local ....

[Article contains additional citation context not shown here]

Payton D. W. et al., "Plan guided reaction", IEEE Transactions on Systems, Man and Cybernetics", vol.20, no.6,pp1370-1382,1990.

....of physical objects whereas behavior based systems usually operate on raw sensory information with limited processing. How to close the gap between planning and behavior based system components is thus an important issue, which has received a considerable amount of attention in the literature [4, 21 25]. Some of the work in hybrid systems deals with integrating planning and reactive behavior based system components. Others, do not exactly integrate the two components, rather they devise new approaches which combine the best of the planning and the behavior based schools. Another direction ....

D. W. Payton, J. K. Rosenblatt, and D. M. Keirsey, "Plan Guided Reaction", IEEE Trans. on Systems, Man, and Cybernetics, Vol. 20, No. 6, pp. 13701382, 1990.

....systems with some hooks to higher level deliberation process either in the concepts or in the architecture. One solution was to provide ways to bias a general reactive system toward the achievement of a goal by imposing constraints [ Drummond, 1989 ] or by exploiting geometric reasoning [ Payton et al. 1990 ] A perhaps more widespread approach sees planning as the goal directed compilation of one instance of some type of reactive machine where the goal, and the corresponding plan, have been hard wired (e.g. Kaelbling, 1988 ] Nilsson, 1992 ] see Figure 1(b) This view of planning is ....

Payton, D. W.; Rosenblatt, J. K.; and Keirsey, D. M. 1990. Plan guided reaction. IEEE Trans. on Systems, Man, and Cybernetics 20(6).

....The fuzzy controller exploits this information, together with the perceptual data, in order to take decisions about the best actions to perform. Viewing plans as sources of information rather than as sequences of instructions provides added flexibility in face of unknown execution contingencies [18, 17, 9]. 2 Behavior schemas The basic building block of the fuzzy controller is a behavior schema, or simply a behavior. A behavior implements an abstract action aimed at achieving a given goal, and corresponds to a motor skill of the agent. Behavioral skills are expressed as preferences over possible ....

D. W. Payton, J. K. Rosenblatt, and D. M. Keirsey. Plan guided reaction. IEEE Trans. on Systems, Man, and Cybernetics, 20(6):1370--1382, 1990.

....[18] are used for navigation of the mobile robots. Navigation of mobile robots is carried out with the use of an environmental map [19] or without creation of such a map [4] Architectural concepts regarding the integration of a high level motion planning with low level reactive behaviours [22] allow to improve the robustness and capabilities of autonomous mobile robots. The reactive behaviours provide a reasonable means to develop a reactive motion control for mobile robots. In our paper the reactive motion control along a preplanned coarse global path is considered. The global path is ....

Payton, D.W., Rosenblatt, J.K., and Keirsey, D.M.: Plan Guided Reaction, IEEE Trans. on Systems, Man and Cybernetics, Vol. 20, No. 6, 1990, pp. 1370-1382.

....rejecting it. In Subsumption higher level layers interface with lower level ones by suppressing the results of the lower level computations and superseding their results. In T R III, higher level layers interfaced with lower level ones by providing input or advice to the lower level layers (cf. Payton90, Agre90] In other words, layers in T RIII provided layers of computational abstraction as well as layers of functionality. Tooth and Rocky III were among the first autonomous robots capable of reliably performing a more complex task than simply moving from place to place, but they had one ....

David Payton, J. Kenneth Rosenblatt and David Keirsey, "Plan-Guided Reaction," IEEE Transactions on Systems, Man and Cybernetics, vol. 20, pp. 1370-1382, 1990.

....the execution of actuator commands from the types of interactions needed between L ctrl close quarters move and L gen propel. A more managable interface across the parameters exists. VI. RELATED RESEARCH There has been a number of approaches to combining planning and reacting for a mobile robot [Firby87, Georgeff87, Rosenschein89, Payton90, Simmons90, Gat91]. None of these approaches have emphasized the determination of what sensor actuator process knowledge is necessary in order to utilize them. Other work has been done in developing schema based computational processes for sensory based robotics [Lyons89, Arkin90] There has also been a great deal ....

David W. Payton, J. Kenneth Rosenblatt, and David M. Keirsey, "Plan Guided Reaction", IEEE Transactions on Systems, Man, and Cybernetics, Vol 20, N. 6, 1990, pp 1370-1382.

....as obstacle detection (SMARTY) and path planning (D ) Each of the modules has its own view of the best action for the robot to take at every step. We use the DAMN architecture for combining recommendations from different modules and for issuing actual driving commands to the robot controller [17][23][26] The DAMN architecture consists of an arbiter which receives steering recommendations from outside modules and combines them into a single driving command. The recommendations are in the form of votes between 1 and 1 for a pre defined set of arcs. A vote of 1 means that the external module ....

Payton, D.W., Rosenblatt, J.K., Keirsey, D.M., "Plan Guided Reaction," IEEE Transactions on Systems Man and Cybernetics, Vol. 20, No. 6, pp. 1370-1382, 1990.

....Pre processing the constructs of a conventional planning model in an effort to achieve timeliness (e.g. 12, 17, 22, 37] 3. Developing formal models of reactive systems (e.g. 22, 24] 4. Combining reactive circuitry with other, more established models, especially conventional planners (e.g. [3, 19, 25, 32, 40]) Much of the work in reactivity has been motivated by application areas where intelligent behavior in the presence of hard temporal deadlines and constraints is desirable. The timing requirements of such problem domains make the techniques of classical planning difficult, if not impossible, to ....

D. W. Payton, J. K. Rosenblatt, and D. M. Keirsey, "Plan Guided Reaction," IEEE Trans. Systems, Man, and Cybernetics, vol. 20, no. 6, pp. 1370--1382, 1990.

....very robustly and reliably. Systems have been developed using a wide variety of techniques which can drive autonomous vehicles on roads [2] 5] 10] 13] avoid obstacles [4] 6] plan safe paths for the vehicle to follow [1] and allow the vehicle to exhibit intelligent, goal directed behavior [9]. There have been attempts to connect these component modules into a robust, comprehensive autonomous driving system. Although many of these attempts have produced good results, there is much work to do in creating a competent, general purpose, autonomous driving system. The foundation of any ....

Payton, D., Rosenblatt, J. and Keirsey, D., "Plan Guided Reaction," IEEE Transactions on Systems Man and Cybernetics, 20(6), 1990, pp. 1370-1382.

....also be applied to the problem of command fusion. Curiously enough, the first proposal of fuzzy command fusion was presented in the literature, in a naive form, by two roboticists who were unaware of fuzzy logic but who were frustrated by the pitfalls of the existing on off arbitration schemas [19]. This proposal has been later re stated in terms of fuzzy logic [8] More generally, fuzzy command fusion can be given a formal setting based on the view of fuzzy logic as a logic of graded preferences [20] Accordingly, we can see each behavior producing unit as an agent expressing preferences ....

D. W. Payton, J. K. Rosenblatt, and D. M. Keirsey. Plan guided reaction. IEEE Trans. on Systems, Man, and Cybernetics, 20(6), 1990.

....plan. action in the navigation plan is represented as a behavior having a context of applicability. This view of plans as sets of situation action rules has been advocated by several authors as a most useful perspective when dealing with situated agents (Suchman, 1987) Schoppers, 1987) (Payton et al. 1990). In our experiment, Flakey was initially at the lower left intersection in the environment shown in the picture. The ultimate navigation goal was to guide Flakey to Room 5. The following behaviors and contexts constitute a plan to reach this goal: IF near(Obstacle) THEN Keep Off IF ....

....behavior at a time has control over effectors. The limitations of this approach have been noticed by several authors who have advocated a weighted form of combination where the outputs of different behaviors are merged and the overall control is the result of a tradeoff computation (Maes, 1989; Payton et al. 1990; Arkin, 1998) Our approach offers a solution to the behavior combination problem that is based on two distinguishing principles: 1. Behaviors are viewed as modules that express goal specific preferences as to the suitability of control actions rather than as low level procedures that compete for ....

[Article contains additional citation context not shown here]

Payton, D. W., Rosenblatt, J. K., and Keirsey, D. M. (1990). Plan guided reaction. IEEE Trans. on Systems, Man, and Cybernetics, 20(6):1370--1382.

.... with planning under uncertainty in realistic situations have led many researchers to question the validity of traditional AI planning techniques for situated agents [ Brooks, 1987 ] and to pro8 pose alternative or, in some cases, complementary approaches (e.g. Firby, 1987; Kaelbling, 1988; Payton et al. 1990 ] Our approach, which utilizes a combination of high level symbolic planning techniques with with low level continuous numerical control, does not require substantial conceptual modifications to the traditional roles of planners and controllers in AI approaches. Rather, our methodology is based ....

....The fuzzy controller utilizes this information, together with perceptual data, to determine the best actions to perform. Considering plans as sources of information rather than as sequences of instructions provides added flexibility when unknown execution contingencies appear [ Suchman, 1987; Payton et al. 1990 ] We will discuss in Section 5 an example where a simple goal regression planner is used to generate sets of control structures that produce complex navigation schemas. 2 4 Self Localization In order to make use of global information about its environment, e.g. a map, an autonomous vehicle ....

Payton, D. W.; Rosenblatt, J. K.; and Keirsey, D. M. 1990. Plan guided reaction. IEEE Trans. on Systems, Man, and Cybernetics 20(6).

....one obstacle at 120 degrees. The dashed line indicates the direction in which the beacon was detected. The solid line indicates the resulting direction command. VII. Comparison to Non Fuzzy Systems Reactive control of mobile robots has also been implemented using a subsumptive behavior approach [3, 4]. As applied to the FuzzBug, this would involve creating separate behaviors for the beacon tracking and the obstacle avoidance functions. The problem with this approach is that one behavior becomes dominant over the other behavior, and the two functions are not fused very well. That is, if the ....

D. W. Payton, J. K. Rosenblatt, and D. M. Keirsey, "Plan guided reaction," IEEE Transactions on Systems, Man, and Cybernetics, vol. 20, no. 6, pp. 1370--1382, November 1990. A. Hardware Schematics and Diagrams B. Code Listings

.... is to adopt a twolevel model: at the upper level, a planner decides a sequence of abstract goals to be achieved, based on the available knowledge; at the lower level, a complex controller achieves these goals while dealing with the environmental contingencies (e.g. McD90,Con92,Gat92,Ark90,PRK90] The controller is complex because it must be able to simultaneously satisfy strategic goals coming from the planner (e.g. going to the end of the corridor) and low level innate goals (e.g. avoiding obstacles on the way) It is the controller s job to produce physical movements that ....

....of this complexity. Path plannning techniques are not incompatible with our behavioral approach: an agent can be equipped with a path planning and following behavior, and combine it with other behaviors when needed. We have implemented such a behavior for Flakey using a method similar to Payton s [PRK90] A gradient field is generated at intervals by a grid search from the goal position to the robot. At each point, the gradient gives the direction the robot should travel in the shortest path to the goal. A path following control schema prefers the directions that are close to those indicated ....

[Article contains additional citation context not shown here]

D. W.Payton, J. K. Rosenblatt, and D. M. Keirsey. Plan guided reaction. IEEE Trans. on Systems, Man, and Cybernetics, 20(6), 1990.

....is insufficient for effective command fusion. Many control systems have been constructed using fuzzy logic [Lee, 1990] and in particular many fuzzy systems have been developed for mobile robot navigation [Saffiotti, 1997] A previous type of actuation space arbiter implemented within DAMN [Payton et al. 1990] operates in a similar manner and in fact has been recast into a fuzzy logic framework and used for control of a mobile robot [Yen and Pfluger, 1992] However, the DAMN actuation arbiter is somewhat more general in that it allows for an arbitrary distribution of votes rather than being restricted ....

David Payton, Julio Rosenblatt, David Keirsey. Plan Guided Reaction. In IEEE Transactions on Systems Man and Cybernetics, 20(6), pp. 1370-1382, 1990.

No context found.

D. W. Payton, J. K. Rosenblatt, and D. M. Keirsey. Plan guided reaction. IEEE Trans. on Systems, Man, and Cybernetics, 20(6):1370--1382, 1990.

No context found.

Payton, D., Rosenblatt, J., Keirsey, D., "Plan Guided Reaction." IEEE Transactions on Systems, Man and Cybernetics, Vol. 20, No. 6, pp. 1370-1382, Nov. 1990.

No context found.

Payton, D., Rosenblatt, J., Keirsey, D., "Plan Guided Reaction", in IEEE Transactions on Systems Man and Cybernetics , Vol. 20, No. 6, pp. 1370-1382, 1990.

No context found.

D. W. Payton, J. Kenneth Rosenblatt, and David M. Keirsey, "Plan Guided Reaction", IEEE Transactions on Systems, Man, and Cybernetics, Vol 20, N. 6, 1990, pp 1370-1382.

First 50 documents

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