Andrea Bonarini. ELF: Learning incomplete fuzzy rule sets for an autonomous robot. In Hans-Jurgen Zimmermann, editor, First European Congress on Fuzzy and Intelligent Technologies { EUFIT'93, volume 1, pages 69-75, Aachen, D, 1993. Verlag der Augustinus Buchhandlung.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....Barrett, David, 437] Bartscht, E. 33] Beer, Randall D. 216] Bennett III, Forest H. 288] Bersano Begey, Tommaso F. 242] Bessi ere, Pierre, 160, 329, 416, 417, 418, 419, 420, 421, 422, 423, 424] Bikdash, M. 296] Biondi, Joelle, 146] Blume, Christian, 102, 370] Bonarini, Andrea, [330] Boone, G. 86] Both, Hans Heinrich, 434] Boudreau, R. 259] Bradshaw, A. 103] Braunstingl, R. 127, 164] Bressgott, W. 33] Brevart, V. 128] Brillowski, K. 260] Brooks, Rodney A. 331] Browne, David, 211] Bruce, Wilker Shane, 289] Buckles, Bill P. 41] Bull, Lawrence, 129, ....

.... 393, 64, 129, 212] distributed, 167] fuzzy, 142, 152] classifiers, 346, 429] co evolution, 241] coding DNA, 316] comparison backpropagation, 132] in control, 145] Levenberg Marquardt, 132] stochastic automata, 101] computational geometry, 48] computer graphics graphs, 23] control, [347, 330, 381, 392, 86, 144, 190] adaptive, 135, 32, 251, 325] architecture, 71] autonomous robots, 180] classifier systems, 167] fuzzy, 430, 79, 80, 82, 12, 434, 149, 164, 191, 199, 218, 233, 32, 252, 256, 272, 438, 325] inverted pendulum, 145] Lyapunov, 244] mobile robot, 163, 204] mobile robots, 64] motion, 24, ....

[Article contains additional citation context not shown here]

Andrea Bonarini. ELF: learning incomplete fuzzy rule sets for an autonomous robot. In ?, editor, Proceedings of EUFIT '93, pages 69--75, Aachen (Germany), ? 1993. ELITE Foundation. y(Dorigo) ga:Bonarini93a.

....M. L. 732] Bergman, A. 11] Beritelli, Francesco, 378] Bernier, J. L. 426] Bersini, Hugues, 672] Bezdek, James C. 484, 132] Bhandari, Dinabandhu, 60, 451] Bianchi, D. 216] Bikdash, M. 634] Billing, G. 107] Blume, H. 768] Bodenhofer, U. 603] Bonarini, Andrea, [458, 164, 321] Bonelli, Pierre, 673] Bonissone, Piero P. 67, 752, 312] Boor, Stefan, 313] Boozarjomehry, R. 602] Born, Joachim, 322] Both, Hans Heinrich, 68] Botta, M. 439] Bowen, James, 561] Bowen, Xu, 600] Bowerman, C. G. D. 628] Boyd, R. 323] Bradford, C. 628] Brandstatter, ....

.... [477] comparison fuzzy approach, 97] fuzzy sets, 756] in control, 89] integer programming, 310] metaheuristics, 310] mutation, 428] computer graphics redering, 138] computer graphics , 384] computer science real time systems, 770] computers PC con guration, 758] control, [672, 443, 350, 360, 321, 690, 347, 349, 696, 140, 419, 281] control adaptive, 485, 489, 376, 730, 503, 745, 205, 210, 670] adaptive force, 717] aircraft, 535] autonomous systems, 95] cart centering, 78] cart pole, 49, 516, 152, 220, 275] chemical reaction, 445] docking a truck, 702] force, 555] fuzzy, 442, 343, 685, 350, 692, 777, ....

[Article contains additional citation context not shown here]

Andrea Bonarini. ELF: learning incomplete fuzzy rule sets for an autonomous robot. In ?, editor, Proceedings of EUFIT '93, pages 69-75, Aachen (Germany), ? 1993. ELITE Foundation. yDorigo ga:Bonarini93a.

....[118] Biethahn, Jorg, 756] Billings, S. A. 287] Biondi, J. 119] Bishop, J. M. 503, 504] Bitterman, Thomas A. 120] Blanchet, Max, 121] Blanton, Jr. Joe L. 122] Bloemer, A. 626] Bluff, K. 123] Blume, Christian, 1043] Bommel, Patrick van, 124, 646] Bonarini, Andrea, [125] Bonelli, Pierre, 126] Bonnet, J erome, 64, 65] Booker, Lashon B. 383] Booker, Peter, 796] Boone, G. 127] 16 Genetic algorithms of 1993 Born, Joachim, 128, 129, 130, 131] Bornholdt, Stefan, 132] Botta, M. 371] Bounds, D. G. 133] Boyd, R. 134] Bradshaw, J. 719] Brandon, ....

.... [251] competition, 817] complexity, 45, 880] compression, 631, 700] computational geometry, 1084] triangulation, 1051] computer graphics, 617] computer networks, 562, 602] computer science viruses, 48] context free grammar induction, 1071] context free grammars, 110] control, [108, 125, 151, 192, 259, 280, 288, 292, 367, 436, 482, 543, 552, 554, 558, 608, 667, 711, 813, 906, 1009, 1017, 1061] control bio, 14] chemical reaction, 559] distributed systems, 511] environmental, 162] exhaust emissions, 728] fuzzy, 1044, 1077] inverted pendulum, 434] nonlinear, 44, 434, 781] pH, 553] PID, 607, 1010] process control, 244, 560] rules extraction, 454] tracking, 905] ....

[Article contains additional citation context not shown here]

Andrea Bonarini. ELF: learning incomplete fuzzy rule sets for an autonomous robot. In ?, editor, Proceedings of EUFIT '93, pages 69--75, Aachen (Germany), ? 1993. ELITE Foundation. y(Dorigo) ga:Bonarini93a.

....similar to another given by an expert, it is necessary to combine and simplify the complete set of rules for obtaining the final set of rules. Finally, the tuning method presented in [Her95a] is applied over the simplified set of rules. Other methods have been proposed under different hypothesis [Bon93, Lee93a, Lee93b, Chw94, Lee94, Hof94, Sat94]. 5 Learning Classifier Systems 5.1 Introduction The man s life is surrounded by many monotonous, difficult and dangerous tasks. Many of them have been eliminated through the developments in autonomous machinery but certain tasks have remained resistant to automation. In particular, tasks that ....

Bonarini, A., ELF: learning incomplete fuzzy rule sets for an autonomous robot. Proc. First European Congress on Fuzzy and Intelligent Technologies, Aachen, 1993, 69-75.

.... Fuzzy Systems There are different aspects on the design of FS and FLCS, as for instance, how to get the set of fuzzy control rules, to decide the number of fuzzy rules, to decide the shape of the membership functions, to tune the fuzzy rules base, etc, in which GA have been successfully applied [Bon93, Cas93, Gey92, 93, Gon93, Her93a, c, Kar91a, b ,c, Kro93, Lee93a, Nom92, Sur93, Tak93, Thr91, Val91a, b]. As it is known, the problem of managing a fuzzy rules base (adquisition, learning, tuning) is of utmost importance in the development of fuzzy systems. In a general learning process we can distinguish different components: 1. A generation method of desirable fuzzy rules able to include the ....

Bonarini, A., ELF: learning incomplete fuzzy rule sets for an autonomous robot. Proc. First European Congress on Fuzzy and Intelligent Technologies, Aachen, 1993, 69-75.

No context found.

Andrea Bonarini. ELF: Learning incomplete fuzzy rule sets for an autonomous robot. In Hans-Jurgen Zimmermann, editor, First European Congress on Fuzzy and Intelligent Technologies { EUFIT'93, volume 1, pages 69-75, Aachen, D, 1993. Verlag der Augustinus Buchhandlung.

....uncertainty in sensors and actuators. The aim of this paper is to discuss the issues concerning learning crisp and fuzzy interval representations when operating with real valued input output; in particular we do not introduce new RL algorithms, presented and compared with other proposals elsewhere [6], 1] 2] 3] 4] nor we discuss about genetics, generalization, or the reinforcement function in our proposal. The considerations wepresentin this paper are general enough to be relevantforany RL algorithm operating on real values. 2 Grids, granularity, and real values The information ....

....step is identical to the evaluation step. A possibilitytokeep a fast reaction, but also a good chance of changing state between two subsequentevaluation steps, is to select an evaluation step larger than the control step. A rst intuition about the importance of this problem was already present in [6] [2] where the term episode was used to name a sequence of control steps at the end of which a reinforcement is given. An even more 4 radical approachistaken by many researchers, who evaluate the performance only when the system reaches a di erentinterval (or fuzzy) state. The di erence with ....

Andrea Bonarini. ELF: Learning incomplete fuzzy rule sets for an autonomous robot. In Hans-Jurgen Zimmermann, editor, First European Congress on Fuzzy and Intelligent Technologies { EUFIT'93,volume 1, pages 69-75, Aachen, D, 1993. Verlag der Augustinus Buchhandlung.

....The extensions presented in this section are only two of the many possible, and we are extensively study many possibilities, in thi framework. IV. ELF and its applications In this section, we summarize the main features of an algorithm (ELF ) that we have independently developed in these years [2], 4] 5] following the principles that we have reported in section III A. Again, this is just an example of an implementation of the methodological approach. The aim of this section is to make some considerations about some of the choices we can take about some components of a LFCS. In ELF, we ....

....The adoption of fuzzy states makes close models (rules) to interact locally, thus propagating the local evolution to the whole system, as new niches are explored. We have applied ELF to many aspects of autonomous agent control, by obtaining satisfying results. We have learnt reactive controllers [2], strategic controllers [7] control modules coordination [3] multi agent coordination. In some cases we have also obtained satisfying controllers in a learning time up to two orders of magnitude smaller than other GA based learning systems [5] A. An example To give an idea of one of the ....

A. Bonarini, ELF: learning incomplete fuzzy rule sets for an autonomous robot, Proceedings of EUFIT '93, ELITE Foundation, Aachen, Germany, pp. 69-75, 1993.

....in the work we are presenting in this paper, since the navigation task is performed in an unknown environment. 3. Fuzzy ARTMAP A Fuzzy ARTMAP neural network (Carpenter et al. 1992) can classify its input according to user defined classes. The input is given by a vector of analogical values in [0,1]. A Fuzzy ARTMAP network is composed of two ART (Carpenter and Grossberg, 1987) modules connected by a MAP module (see figure 2) The ARTa module receives the set of values to be classified (a) and produces a classification (y a ) based on a similarity criterion, and a resonance mechanism that ....

A. Bonarini, 1993, ELF: learning incomplete fuzzy rule sets for an autonomous robot, Proceedings of EUFIT '93, ELITE Foundation, Aachen, Germany, pp. 69-75.

No context found.

Bonarini A., 1993, ELF: learning incomplete fuzzy rule sets for an autonomous robot, Proc. of EUFIT '93, ELITE Foundation, Agentchen, Germany, pp. 69-75.

....rules that produce a robust FLC have different antecedents. Rules with the same antecedent usually compete with each other, since they propose different actions for the same state. Therefore, a possible solution to this problem is to have a population of rules partitioned into sub populations [2]. Each sub population contains rules that have the same antecedents, thus they match the same fuzzy state. Rules in a subpopulation compete with each other. Rules belonging to different sub populations cooperate to achieve robust behaviors. This is different both from what proposed by people ....

....many real world environments. Moreover, the verification of the correctness of a reinforcement program is still an open problem [3] 9] We have found support to this opinion about the difficulty to evaluate the correctness of a reinforcement program in different experiments with autonomous agents [2][3] 5] and other fuzzy control applications, such as: the cart pole balancing [14] 17] the spacecraft autonomous rendez vous [15] and the truck backing up [17] We have found that it is common to have critical states that are not considered by reinforcement programs and that naturally arise to ....

A Bonarini, ELF: learning incomplete fuzzy rule sets for an autonomous robot, Proc. of EUFIT '93, ELITE Foundation, Aachen, Germany, pp. 69-75, 1993.

....a successful design practice, according to the principle of problem decomposition. Among the possible implementations for behaviors, the most common are: finite state machines, following the original Brooks s approach [11] 19] 18] classifier systems [13] 23] and fuzzy rules [21] [2], 4] 6] The combination of the different basic behaviors is often obtained through the application of inhibition mechanisms, in some cases integrated in the subsumption architecture [11] 19] 18] The adoption of a fuzzy logic representation makes it possible also other forms of ....

....the output of basic behaviors. In this paper, we present S ELF (Symbolic ELF) a reinforcement learning [17] system that learns to coordinate pre defined basic behaviors, by identifying the best contexts for each of them. We have developed S ELF from ELF (Evolutionary Learning of Fuzzy rules) [2], 4] a system that we have successfully adopted in the past to learn fuzzy behaviors (i.e. behaviors implemented by fuzzy rules) and their coordination [3] 6] 8] S ELF learns the context of activation for each available, basic behavior. It works on contexts described by logical ....

A. Bonarini, ELF: learning incomplete fuzzy rule sets for an autonomous robot, Proc. of EUFIT '93, ELITE Foundation, Aachen, Germany, pp. 69-75, 1993.

No context found.

Bonarini A., 1993, ELF: learning incomplete fuzzy rule sets for an autonomous robot, Proc. of EUFIT '93, ELITE Foundation, Agentchen, Germany, pp. 69-75.

....in selecting a limited set of variables and values, such that the search space is small. Fuzzy logic makes this possible without loosing the precision of data available from sensors and actuators. We have exploited with success this possibility in ELF (Evolutionary Learning of Fuzzy rules)[2][4] Another approach to increase the convergence speed of a reinforcement learning algorithm is viable when the global task can be decomposed into simpler ones, and behaviors performing the simpler tasks cooperate to obtain a good performance also in the global task. This approach has been taken ....

....travel at a maximum speed of 0.2 m s. It has a 20 cm. high turret that holds infrared emitters and implements a 360 wide beacon. 3 . DESIGNING THE CONTROL ARCHITECTURE To learn the above mentioned task, we adopt a control architecture based on behaviors [6] and implemented by sets of fuzzy rules [2], 4] Since in this case we want to cope also with dynamic aspects of the environment, we have decided to have behaviors that explicitly consider these aspects. For instance, for the prey following behavior, we have decided to have two interacting control systems: a basic, reactive system that ....

[Article contains additional citation context not shown here]

A Bonarini, (1993). ELF: learning incomplete fuzzy rule sets for an autonomous robot, Proc. of EUFIT '93, ELITE Foundation, Aachen, Germany, pp. 69-75.

....(and, in general, Evolutionary Learning (EL) algorithms) to learn Fuzzy Logic Controllers (FLC) In particular we focus on a real time learning application such as the control of autonomous agents. We have developed an evolutionary learning algorithm (ELF Evolutionary Learning of Fuzzy rules)[2][3] 6] that we have successfully applied to learn FLCs for simulated and real robots. In this paper, we discuss problems related to the application of Evolutionary Learning algorithms to learn FLCs. In particular, we focus on the problem of learning chains of rules when the reinforcement is not ....

....consider Q Learning [16] as a successful solution of the problem. With the increasing interest about Fuzzy Logic applications, some researchers have investigated the possibility of extending Q Learning with some Fuzzy Logic features [1] 9] Fuzzy Q Learning) Independently, we have implemented ELF [2], whose reinforcement distribution algorithm is oriented to achieve results analogous to those provided by Fuzzy Q Learning. In this paper we discuss delayed reinforcement, Q Learning and its fuzzy extensions. Moreover, we mention other approaches concerning learning FLCs with delayed ....

[Article contains additional citation context not shown here]

A Bonarini, ELF: learning incomplete fuzzy rule sets for an autonomous robot, Proc. of EUFIT '93, ELITE Foundation, Aachen, Germany, pp. 69-75, 1993.

No context found.

Bonarini A., 1993, ELF: learning incomplete fuzzy rule sets for an autonomous robot, Proc. of EUFIT '93, ELITE Foundation, Aachen, Germany, pp. 69-75.

....although it is based on few linguistic values (Saffiotti, Konolige, and Ruspini, 1995) Bonarini, 1996a) We discuss the use of fuzzy sets to represent the variables of the control system in Section 2. We have successfully adopted fuzzy models in ELF (Evolutionary Learning of Fuzzy rules) (Bonarini, 1993), Bonarini, 1994a) Bonarini, 1996a) the learning system that we have also used in the experimental activity presented in this paper. In Section 3, we present ELF and discuss its main features, including its niche based partitioning of the population, which reduces competition among different ....

Bonarini, A. (1993). ELF: learning incomplete fuzzy rule sets for an autonomous robot. Proceedings of the EUFIT '93. Aachen, D: ELITE Foundation.

No context found.

Bonarini A., 1993, ELF: learning incomplete fuzzy rule sets for an autonomous robot, to appear in Proc. of EUFIT '93, ELITE Foundation, Aachen, Germany.

No context found.

Bonarini A., 1993, ELF: learning incomplete fuzzy rule sets for an autonomous robot, Proc. of EUFIT '93, ELITE Foundation, Agentchen, Germany, pp. 69-75.

No context found.

A. Bonarini. ELF: Learning incomplete fuzzy rule sets for an autonomous robot. In Hans-Jurgen Zimmermann, editor, First European Congress on Fuzzy and Intelligent Technologies { EUFIT'93, volume 1, pages 69-75, Aachen, September 1993. Verlag der Augustinus Buchhandlung.

No context found.

Andrea Bonarini. ELF: Learning Incomplete Fuzzy Rule Sets for an Autonomous Robot. In Hans-Jurgen Zimmermann, editor, First European Congress on Fuzzy and Intelligent Technologies { EUFIT'93, volume 1, pages 69-75, Aachen, D, September 1993. Verlag der Augustinus Buchhandlung.

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