W-P. Lee et al. Learning complex robot behaviours by evolutionary computing with task decomposition. In Andreas Brink and John Demiris, editors, Learning Robots: Proc. of 6th European Workshop, EWLR-6 Brighton. Springer, 1997.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....it is possible to evolve multiplier circuits larger then those evolved earlier. 1 Introduction For the recent years, evolvable hardware (EHW) has become an important scheme for automatic circuit design. However, still there lack schemes to overcome the limitation in the chromosome string length [1, 2]. A long string is required for representing a complex system. However, a larger number of generations are required by genetic algorithms (GA) as the string length increases. This often makes the search space too large and explains why only small circuits have been evolvable so far. Thus, work has ....

W-P. Lee, J. Hallam, and H.H. Lund. Learning complex robot behaviours by evolutionary computing with task decomposition. In A. Birk and J. Demiris, editors, Learning Robots: Proc. of 6th European Workshop, EWLR-6 Brighton, volume 1545 of Lecture Notes in Artificial Intelligence, pages 155--172. Springer-Verlag, 1997.

....actions. At each 100 ms time step, control is propagated down through the tree to one of the 36 primitives, which then has full control of the robot for the duration of the time step. The architecture is based on the task decomposition approach described by W. Lee, J. Hallam and H. H. Lurid [LHL97] Two types of arbitrators were used, implementing sequential and reactive arbitration, both illustrated in figure 5. Both arbitrators uses a set of conditions to arbitrate between six sub modules, and transfer control to only one of them. The conditions are implemented as fixed size boolean ....

Wei-Po Lee, John Hallam, and Henrik Hautop Lund. Learning complex robot behaviours by evolutionary approaches. 6th European Workshop on Learning Robots, EWLR-6, aug 1997.

....evolution with higher level behavior primitives. The primitives can be obtained by evolving controllers to perform the basic actions of the robot, such as turning and moving forwards. Providing these higher level primitives for the evolutionary search should make it easier to find a good solution. [7] Rationale: If none of the solutions generated by the evolutionary algorithm come close to performing the task, and thus all get the same low fitness score, the selection mechanism doesn t work. The system must be changed, so that selection pressure can kick in. By changing a combination of the ....

Wei-Po Lee, John Hallam, and Henrik Hautop Lund. Learning complex robot behaviours by evolutionary approaches. 6th European Workshop on Learning Robots, EWLR-6, aug 1997.

....one complete circuit. The simulation indicates a similar performance as artificial neural network but since the EHW controller requires a much smaller hardware it is to be preferred. One of the main problems in evolving hardware systems seems to be the limitation in the chromosome string length [4, 5]. A long string is normally required for representing a complex system. However, a larger number of generations is required by genetic algorithms (GA) as the string increases. This often makes the search space too large. Thus, work has been undertaken to try to diminish this limitation. Various ....

W-P. Lee et al. Learning complex robot behaviours by evolutionary computing with task decomposition. In Andreas Brink and John Demiris, editors, Learning Robots: Proc. of 6th European Workshop, EWLR-6 Brighton. Springer, 1997.

....for a certain time. 3.5 Evolving Complex Systems The work described in Section 2 is mainly based on circuits with a limited number of building blocks. Thus, the applications have limited complexity. To solve more complex applications, the limitation in the chromosome string length must be solved [26, 27]. A long string is required for representing a complex system. However, a larger number of evolutionary generations are required as the string increases. This often makes the search space too large and explaines why only small circuits have been evolvable so far. Thus, work has been undertaken to ....

W-P. Lee et al. Learning complex robot behaviours by evolutionary computing with task decomposition. In Andreas Brink and John Demiris, editors, Learning Robots: Proc. of 6th European Workshop, EWLR-6 Brighton. Springer, 1997.

....Introduction Evolvable hardware (EHW) has recently been introduced as a new scheme for designing systems for real world applications [17] So far the number of applications is limited. One of the main problems in evolving hardware systems seems to be the limitation in the chromosome string length [11, 19]. A long string is normally required for representing a complex system. However, a larger number of generations is required by genetic algorithms (GA) as the string increases. Thus, work has been undertaken to try to diminish this limitation. Various experiments on speeding up the GA computation ....

....are divided without human intervention [5] Task decomposition for robot controllers have been proposed in several different ways. Chavas et al. [4] have developed a two stage system using incremental evolution for a neural network based controller. Another method, proposed by Lee et al. [11], evolves a distributed control architecture. Every module only deals with the sensory information directly related to its particular need. Incremental evolution for EHW was first introduced in [16] for a character recognition system. The approach is a divide and conquer on the evolution of the ....

W.-P. Lee et al. Learning complex robot behaviours by evolutionary computing with task decomposition. In A. Brink and J. Demiris, editors, Learning Robots: Proc. of 6th European Workshop, EWLR-6 Brighton. Springer, 1997.

....be described by a probability distribution. If this probability distribution changes over time we call the fitness landscape dynamic. Dynamic fitness landscapes are typical phenomena encountered in the field of evolutionary robotics especially in on line evolutionary experiments on real robots [5, 10, 12, 13, 16]. Another domain dealing with dynamical fitness landscapes is co evolution [6, 11] where the fitness of an individual depends on the phenotype of the individuals in the current population. This is qualitatively different to the robotic experiments, because here the fitness landscape depends on the ....

Wei-Po Lee, John Hallam, and Henrik Hautop Lund. Learning complex robot behaviours by evolutionary approaches. In 6th European Workshop on Learning Robots, pages 42--51, Brighton, UK, 1-2 August 1997.

.... However, the fact that the great majority of controllers and behaviors that have thus been generated are very simple, together with the difficulties encountered when more complex controllers and behaviors were sought [11,20] led us to suspect that so called incremental approaches [4,12,23] should necessarily be used in conjunction with indirect encoding schemes in more realistic applications. In other words, according to such a strategy, appropriate controllers and behaviors should be evolved and developed through successive stages in which good solutions to a simpler version of a ....

W. Lee, J. Hallam and H. Lund, Learning complex robot behaviours by evolutionary approaches, in Proceedings of the 6th european workshop on learning robots, Brighton, (1997)

.... 11, 28, 29] However, the fact that the great majority of controllers and behaviors that have thus been generated are very simple, together with the diculties encountered when more complex controllers and behaviors were 1 sought [11, 20] led us to suspect that so called incremental approaches [4, 12, 23] should necessarily be used in conjunction with indirect encoding schemes in more realistic applications. In other words, according to such a strategy, appropriate controllers and behaviors should be evolved and developed through successive stages in which good solutions to a simpler version of a ....

W. Lee, J. Hallam and H. Lund, Learning complex robot behaviours by evolutionary approaches, in Proceedings of the 6th european workshop on learning robots, Brighton, (1997)

.... it will be to devise fitness functions likely to automatically select complex behaviors, even if so called incremental approaches according to which the overall behavior is decomposed into simpler behavioral primitives that are successively evolved and combined together (e.g. 30] 31] [34]) seem to be helpful. Likewise, if indirect coding affords the evolutionary process the possibility of exploring smaller search spaces than direct coding does, it is likely that devising and adjusting the corresponding genetic operators e.g. mutations and crossovers will prove to be much ....

Lee, W.P., Hallam, J. and Lund, H.H. Learning Complex Robot Behaviours by evolutionary Approaches. Proceedings of the 6th European Workshop on Learning Robots. Brighton. 1997.

....Evolvable hardware (EHW) has recently been introduced as a new scheme for designing systems for real world applications. So far the number of applications is highly limited. One of the main problems in evolving hardware systems seems to be the limitation in the chromosome string length [7]. A long string is required for representing a complex system. However, a larger number of generations are required by genetic algorithms (GA) as the string increases. Thus, work has been undertaken to try to diminish this limitation. Various experiments on speeding up the GA computation have been ....

W-P. Lee et al. Learning complex robot behaviours by evolutionary computing with task decomposition. In Andreas Brink and John Demiris, editors, Learning Robots: Proc. of 6th European Workshop, EWLR-6 Brighton. Springer, 1997.

No context found.

W-P. Lee et al. Learning complex robot behaviours by evolutionary computing with task decomposition. In Andreas Brink and John Demiris, editors, Learning Robots: Proc. of 6th European Workshop, EWLR-6 Brighton. Springer, 1997.

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