J-A. Meyer and A. Guillot. From SAB90 to SAB94: Four years of animat research. In D. Cli, P. Husbands, J-A. Meyer, and S.K. Wilson, editors, Proceedings of the Third International Conference on Simulation of Adaptive Behavior. MIT Press, Cambridge, MA, 1994.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....emerge and both the best and average tness in the population increase until a plateau is reached. Current EAA studies have been able to successfully evolve arti cial networks of several dozen neurons and hundreds of synapses, controlling agents performing non trivial behavioral tasks (see [6, 7, 8, 9, 10] for recent reviews) These networks are less biased than conventional neural networks used in neuroscience modeling as their architecture is not pre designed in many cases. They are 4 0 5000 10000 15000 20000 25000 30000 0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Generation Fitness ....

J-A. Meyer and A. Guillot. From SAB90 to SAB94: Four years of animat research. In D. Cli, P. Husbands, J-A. Meyer, and S.K. Wilson, editors, Proceedings of the Third International Conference on Simulation of Adaptive Behavior. MIT Press, Cambridge, MA, 1994.

....composing the agent s controller network. 1 1 Introduction Many Evolutionary Autonomous Agent (EAA) models use neural networks to control their behavior. Such EAA neurocontrollers are evolved via genetic algorithms from a population of genomes undergoing natural selection and variation (see [21, 32, 9, 8] for a review ) The majority of these studies employ direct genotype to phenotype encodings, where the magnitude of each synapse is encoded by a speci cally designated gene. These direct encodings scale quadratically with network size and hence are hypothesized to be inadequate for solving ....

J.A. Meyer and A. Guilliot. From SAB90 to SAB94:four years of animat research. Proceedings of the third international Conference on Simulation of Adaptive Behavior.,ed D.Cli and P.Husbands and J.A.Meyer and S.K.Wilson MIT Press,Cambridge,MA, 1994.

....structure on which genetic operators were defined. The EASEA compiler converted the specification into a compilable C source file using GAlib. 1. 2 The artificial ant problem, animats and low level image processing The Artificial Ant problem [18] 15] considers the task of navigating an animat [25][28] gathering food scattered irregularly (e.g. using the Santa Fe trail) over a twodimensional domain represented as a binary image. This problem, known as containing multiple levels of deception (GA hard) 23] has often been used as a benchmark for Genetic Programming (GP) In the first part of ....

....The input is a natural grey level image: the animat s task is to find as many contour pixels as possible, using a minimal path length. Again, the output is a binary image showing the pixels the animat has detected. The behavioural aspects of several different animat types have been studied [25] and linked to real world behaviours in Biology [19] 27] 2] but as far as we know, only few applications to real image processing have been published. See [16,17] for crack detection with image exploring agents . 2 Implementing a food foraging process using EASEA 2.1 Animat functions The ....

[Article contains additional citation context not shown here]

J. A. Meyer, A. Guillot, From SAB90 to SAB94: Four Years of Animat Research, Proceedings of Third International Conference on Simulation of Adaptive Behavior. Brighton, England, 1994.

....[61] Dewdney, A. K. 62] Donnart, Jean Yves, 11] Dyer, Michael G. 68, 69, 73, 74] Emmeche, Claus, 12, 29, 65] Flower, Joe, 49] Flowers, Margot, 73, 74] Forrest, Stephanie, 21, 34, 66] Furuya, T. 105] Gallagher, John C. 59] Garis, Hugo de, 26, 27, 106, 107, 108] Guillot, Agn es, [13, 19, 25] Heudin, J. C. 40] Horn, Jeffrey, 71] Hyvonen, Eero, 30, 41] Inayoshi, H. 14] Jefferson, David R. 73, 74, 85, 86] Johnson, R. C. 83] Juille, H. 42] Kimura, M. 43] Korf, Richard, 73, 74] Langton, Christopher G. 76] Leino, Raili, 16] Levy, S. 77] Lohn, Jason D. 32] ....

....[43] Korf, Richard, 73, 74] Langton, Christopher G. 76] Leino, Raili, 16] Levy, S. 77] Lohn, Jason D. 32] Lopez, Luis R. 84] Maeshiro, T. 43] Maley, Carlo C. 78] Mancuso, S. 47] Mange, D. 79] Martin, Jose Luis FernandezVillanacas, 33] McNeil, S. 38] Meyer, Jean Arcady, [11, 13, 15, 19, 20, 25] Miller, Geoffrey F. 67] Mitchell, Melanie, 21, 34, 66] Nagahashi, Hiroshi, 35] Nagahashi, H. 28] Niwa, Akihiro, 35] Ohira, T. 44] Ono, N. 44] Pargellis, A. 50] Parisi, D. 47] Paukku, Timo, 36, 45, 80, 81] Pietilainen, Kimmo, 37] Putnam, Jeffrey, 82] Rahmani, A. T. 44] ....

[Article contains additional citation context not shown here]

Jean-Arcady Meyer and Agn'es Guillot. From SAB90 to SAB94: Four years of animat research. In David Cliff, Philip Husbands, and Jean-Arcady Meyer, editors, From Animals to Animats 3. Proceedings of the Third International Conference on Simulation of Adaptive Behavior, page ?, ?, ? 1994. MIT Press, Cambridge, MA. y(Meyer) ga94dMeyer.

....mechanisms of heredity and adaptation was required. A highly modified classifier system [BGH90] was developed as the adaptive engine for each agent. The classifier as conceived here was not intended to perform complex machine learning tasks it has been directed towards in other work, e.g. Wil90] [MG94]. It was intended purely as a mechanism of basic adaptation. This simplification was necessary in order to run simulations containing large numbers of autonomous agents over very long virtual time periods. The classifier system (CS) employed follows the Michigan style [Hol92] which has been ....

J. A. Mayer and A. Guillot. From sab90 to sab94: Four years of animat research. In Cliff, Husbands, Meyer, and Wilson, editors, From animals to animats 3. Proceedings of the third international conference on simulation of adaptive behaviour. MIT Press, 1994.

....of populations and communities. The animat approach [20] has established itself as fertile ground for investigating adaptive behaviours in real, or more usually, simulated environments. Various mechanisms for adaptive functionality have been employed and their applications are well documented [14]. In common with much animat research, that described here is directed at the study of adaptation in a well defined context, namely a class of ecologically realistic herbivore environments. Other work devoted to similar systems has tended to be targeted at behavioural adaptation in either single ....

J. A. Meyer and A. Guillot. From sab90 to sab94: Four years of animat research. In Cliff, Husbands, Meyer, and Wilson, editors, From animals to animats 3. Proceedings of the third international conference on simulation of adaptive behaviour. MIT Press, 1994.

.... individuals and consequent differential benefits are central [DC89] The mechanism of adaptation used is a modified Michigan style classifier system [Hol90] This classifier system [Dev96] has not been designed to effect the sort of complex machine learning tasks that others have been used for [MG94], but rather as a vehicle for adaptations within large populations of competing, co evolving agents. The overall classifier system is organised along established lines [BGH90] with the exception of the rule set. This is partitioned into two sets (S and S , see Fig. 2) though for the purposes of ....

J. A. Mayer and A. Guillot. From sab90 to sab94: Four years of animat research. In Cliff, Husbands, Meyer, and Wilson, editors, From animals to animats 3. Proceedings of the third international conference on simulation of adaptive behaviour. MIT Press, 1994.

....term, intermediate term, and ultimate goals. In the short term it is the discovery and exploration . of architectures and working principles that allow a real animal, a simulated animal, or a robot to exhibit a behavior that solves a specific problem of adaptation in a specific environment. (Meyer and Guillot, 1994, p. 7) In the intermediate term, it is the generalization of this knowledge in order to better understand the relation between architectures, working principles and adaptive performance vis vis different types of environments. The ultimate goal, then, is to understand the adaptive value and ....

Meyer, and S.W. Wilson (eds.). From animals to animats 1. Proc. SAB'90, 2-14. Meyer, J.-A., and Guillot, A. (1994). From SAB90 to SAB94: four years of animat research. In: D. Cliff, P.

....on adaptive behavior in animats. It summarizes the current stateof the art and outlines directions for possible progress. 1. Introduction In the proceedings of SAB94, we published a review of signi cant research on adaptive behavior in animats since the rst SAB conference, held in 1990 (MEYE94) This review summarized the state of the art, insofar as the proceedings of three dedicated conferences could help delineate it. Now that three other SAB conferences have been held, we considered that it would be useful to update that earlier review, in order to assess the corresponding ....

....whose authors seem convinced that this approach will scale with more complex behaviors and survival problems. Likewise, although Aubin s viability theory has just been introduced to the SAB community (AUBI00) it appears in our opinion to be too closely related to animat concerns (see, e.g. MEYE94) for not soon start guiding the intuition of animat designers. There is therefore de nitely some hope that such e orts to bring mathematics and their deductive power into the empiricism of animat research will sooner or later generate the sort of stability theorem or convergence proof that the ....

[Article contains additional citation context not shown here]

J.A. Meyer and A. Guillot. From SAB90 to SAB94: Four Years of Animat Research. In [SAB94].

....Navigation abilities are also very important for mobile robots. But current robots have a limited set of possible behaviors and although they are able to move efficiently through a cluttered environment and to avoid obstacles, their navigation systems tend to be specialized and brittle [78]. Navigation is thus one of the most elaborate tasks that current mobile robots try to accomplish. Animals, on the other hand, can be very good at navigating. The paths they take may be suboptimal in a mathematical sense, but they are rapidly selected, flexible and the resulting navigation ....

....will not be able to reach the goal unless it can wander a little, recognize another place, and use the same local response strategies there (Fig. 2) Trullier et al. 7 Progress in Neurobiology The navigation system uses a mapping from place and goal to directed actions, a type of action model [6, 78, 118]. We will call such mappings place goal action associations . X X X X X X goal X X place local reference direction wandering behavior trajectory between successive choice points X new obstacle direction to next subgoal Figure 2: The place recognition triggered response strategy ....

[Article contains additional citation context not shown here]

J.-A. Meyer and A. Guillot. From SAB90 to SAB94: Four years of animat research. In Cliff et al. [29]. Trullier et al. 85 Progress in Neurobiology

....can be expected to help in understanding how real animals manage to survive in real worlds. On the other hand, the corresponding mechanisms and working principles may prove to be effective in devising truly autonomous and adaptive robots. The literature on animats falls into four broad categories [GUIL94, MEYE91b, MEYE94], according to whether it concerns animats whose control architectures have been programmed by a human designer, or animats whose behavioral plasticity is due to biologically inspired automatic processes like the processes of learning, evolution or development. In order to illustrate the ....

.... used as internal neurons which process visual information (Figure 8) 5 Behavioral Development If learning and evolution have already often been used for the automatic design of control architectures of animats, such is not the case with development a point stressed by Meyer and Guillot [MEYE94]. However, a few applications combining development, evolution and learning have recently been published [KODJ94] The work of Nolfi and Parisi [NOLF91] for instance, is concerned with the evolution of animats that consume food randomly distributed within a simple 2 D environment. Each animat is ....

[Article contains additional citation context not shown here]

J.A. Meyer and A. Guillot. From SAB90 to SAB94 : Four Years of Animat Research. In [?].

....control architecture is fixed by a human designer. However, because the field of animat research still lacks any systematic comparisons where a specific architecture would be confronted with various survival problems or where various architectures would have to cope with the same environment [4, 5, 6] such a procedure relies more on the designer s intuition or technical idiosyncrasies than on basic principles. Actually, several researchers [7, 8] even doubt that such principles will ever prove to be efficient for designing truly autonomous systems, i.e. systems that have to survive in ....

....interact in complex ways, which are still not fully understood. Be that as it may, because the study of animats is grounded in biology, it is not surprising that several research efforts have already let an animat s control architecture evolve, or learn, or simultaneously both evolve and learn [4, 5, 6]. This paper will focus on the evolutionary design of neural networks, a particular class of control architectures liable to exhibit a wide variety of dynamic behaviors and that may prove to be particularly suitable for automatic design approaches because the low level, sub symbolic primitives ....

J.-A. Meyer and A. Guillot. From SAB90 to SAB94: Four years of animat research, in D. Cliff, P. Husbands, J.-A. Meyer, and S. W. Wilson, editors, From Animals to Animats 3. Proceedings of the Third International Conference on Simulation of Adaptive Behavior. The MIT Press/Bradford Books, Cambridge, MA, 1994.

.... so called animats, i.e. simulated animals or real robots, whose structure and functionalities are as much inspired from those of animals as possible, and which exhibit some variety of adaptive behavior [25, 24, 7, 15] Thus, animats can evolve, develop, learn, memorize, plan and communicate [22, 23, 21]. Like animals, instead of being passive reflexive devices, animats often are active data processors that seek useful information in their environment, encode it into internal representations of objects and causal relations, and use these representations for their own benefit in flexible and ....

Meyer, J.-A. and Guillot, A. (1994): From SAB90 to SAB94: Four years of animat research. In Cliff et al. [7].

....process inspired from biology and involving the three main adaptive processes characteristic of living systems, i.e. the processes of development, learning and evolution. Although learning and evolution have already often been used for the automatic design of control architectures in animats [MEYE91b, MEYE94], such does not happen to be the case with the process of development, a point stressed and regretted by Meyer and Guillot [MEYE94] However, a few such applications which combine development, evolution and, possibly, learning have recently been published. They will be described in the ....

....development, learning and evolution. Although learning and evolution have already often been used for the automatic design of control architectures in animats [MEYE91b, MEYE94] such does not happen to be the case with the process of development, a point stressed and regretted by Meyer and Guillot [MEYE94]. However, a few such applications which combine development, evolution and, possibly, learning have recently been published. They will be described in the remainder of this paper, which will close with a discussion of the foreseeable potentialities of such approaches. 2 Boers and Kuiper The ....

J.A. Meyer and A. Guillot. From SAB90 to SAB94 : Four Years of Animat Research. In [CLIF94].

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