Cliff, D., Husbands, P. & Harvey, I. "Analysis of evolved sensory-motor controllers." 2nd European Conference on Artificial Life, Brussels. 1992  Home/Search   Document Details and Download   Summary   Related Articles   Check

....e.g. in artificial life. In the literature we can find several approaches to this problem, the most promising of which seem to be: Neural Networks This is undoubtedly the architecture more often chosen (and more strongly defended) to be evolved as the control system for autonomous agents [Cliff92] [Floreano94] Harvey93] The topology of the networks, however, varies substantially between approaches, and so does what is really evolved: connections weights; weights and connections; weights, connections, and number of neurones, etc. Programs Several authors propose the use of ....

D. Cliff, P. Husbands and I. Harvey, "Analysis of Evolved Sensory-Motor Controllers", Cognitive Science Research Paper, Serial N CSRP 264, 1992.

....a solution from a population of individuals using the principles of natural selection instead of more complex engineering techniques. Evolutionary approaches to agent synthesis can be divided in three main groups according with the representation used for the individuals: Neural networks [4, 8, 10]. Rule based systems [6, 7, 9] Computer programs [1, 2] The choice of the most appropriate controller architecture for autonomous agents is the center of an ongoing discussion [2, 15] which will probably never end. But from it we can conclude that the use of very specific representations ....

D. Cliff, P. Husbands and I. Harvey, "Analysis of Evolved Sensory-Motor Controllers ", Cognitive Science Research Paper, Serial N CSRP 264, 1992.

....Lai, 256] Chisholm, Calum J. A. 959] Chockalingam, T. 60] Chow, C. R. 176] Chow, Keith Hung Kei, 631] Chu, C. H. 176] Chu, Po Hsiang, 780] Chua, Leon O. 177] Cinkosky, M. J. 276] Ciuffolini, D. 724] Clark, David E. 373] Clark, T. 510, 511] Cliff, David T. [436, 437, 438, 439, 440, 441, 442, 443, 446, 447, 449, 450] Cluitmans, L. J. M. 814] Cobb, Helen G. 388] Cohen, Paul R. 62, 63] Colin, Andrew, 178] Collard, Philippe, 179] Collins, Steve M. 887, 888] Colombetti, Marco, 233] Colorni, Alberto, 9, 231] Comellas, F. 180] Cook, A. 239] Cook, Diane J. 194] Cooley, D. 181] ....

....Georges, 376, 378] Harper, T. R. 558, 559] Harris, G. 826, 852] Harris, R. 252, 253, 254] Harris, S. 883] Harris, Stephen P. 161, 162] Harrison, R. F. 433, 434] Hartfelder, Michael, 317] Hartke, Bernd, 435] Hartley, Stephen J. 586] Hartmann, Uwe, 1021] Harvey, Inman, [436, 437, 438, 439, 440, 441, 442, 443, 444, 445, 446, 447, 448, 449, 450] Hase, H. 404] Hasegawa, Yoshishige, 995] Hashier, David J. 451] Hashimoto, Y. 974] Hassoun, Mohamad H. 452, 453, 454, 455] Hatcher, W. J. 558, 559] Hatjimihail, Aristides T. 456, 457] Haupt, R. L. 839] Haupt, Randy L. 697] Hayashi, Yoichi, 139] Hazout, Serge, 924, ....

[Article contains additional citation context not shown here]

David T. Cliff, Philip Husbands, and Inman Harvey. Analysis of evolved sensory-motor controllers. In ?

.... of Cambridge, 53] University of Durham, 351] University of Granada, 181, 199, 200, 202] University of Illinois at Urbana Champaign, 148, 150, 177, 183, 195, 317, 318, 319, 338, 344] University of Nebraska Lincoln, 85, 282] University of Strathclyde, 345, 346, 347] University of Sussex, [114, 117, 119, 123, 320, 321, 322, 323, 324, 325, 326, 327, 328, 329, 331] University of Vaasa, 103, 104, 105, 106, 139, 191, 241, 255, 256, 294, 375, 376, 377, 378, 379, 380, 381, 382, 383, 384, 385, 386, 387, 388, 389, 390, 391, 392, 393, 394, 395, 396, 397, 398, 399, 400, 401, 402, 403, 404, 405, 406, 407, 408, 409, 410, 411, 412, 413, 414, 415, 416, 417, 418, 419] ....

....Cant u Paz, Erick, 150] Caruana, Rich, 44, 45] Carvalho, Luis, 26] Castro, J. L. 333] Cejtin, H. 270] Chalmers, Alan G. 180, 211] Chen, J. R. 74] Chen, Shu Heng, 275] Cheng, Runwei, 91] Chiva, Emmanual, 113] Chorafas, Dimitris N. 12] Chu, Chee Hung H. 112] Cliff, David T. [114, 117, 119, 151, 322, 323, 325, 326, 327, 328, 329, 331] Cockcroft, Victor, 208] Collins, Robert James, 353] Colombetti, Marco, 310, 313] Cootes, T. F. 35] Cord on, Oscar, 152, 181, 192, 199, 203, 212, 235, 245, 248, 250, 252] Coveney, Peter V. 96] Crutchfield, James P. 168, 342, 343] Cziko, Gary, 32] Das, Rajarshi, 28, 168] Dasgupta, ....

[Article contains additional citation context not shown here]

David T. Cliff, Philip Husbands, and Inman Harvey. Analysis of evolved sensory-motor controllers. Technical Report CSRP264, University of Sussex, School of Cognitive and Computing Sciences, 1992. (also as [439]; ftp.cogs.susx.ac.uk: /pub/reports/csrp/ csrp264.ps.Z) Key: ga:Harvey92e.

.... 219, 6] University of Durham, 526] University of East Anglia, 234] University of Illinois at Urbana Champaign, 249, 250, 251, 252, 253, 254, 3, 256, 257, 458] University of Louisville, 705] University of Maryland, 507, 508] University of Michigan, 81, 82, 83] University of Sussex, [298, 299, 301, 302, 303, 304, 305, 306, 307] Universitat Gottingen, 528] Universitat Osnabruck, 412] total 81 reports in 45 institutes 4.5 Patents The following list contains the names of the patents of genetic algorithms of 1992. The list is arranged in alphabetical order by the name of the patent. A non linear genetic algorithms ....

....R. G. 123] Cesare, Mark Anthony, 124] Chakraborty, Uday Kumar, 138] Chan, Shu Park, 600, 601, 602, 603] Chang, Ben, 180] Chen, Qi, 125] Chipperfield, A. J. 193] Chockalingam, T. 62] Choi, H. S. 425] Chung, C. H. 366] Clark, James H. 248, 263] Cliff, David T. [301, 302, 304, 305, 306, 307] Cline, D. D. 557, 558] Cluitmans, L. J. M. 126] Authors 13 Cobb, Helen G. 127] Coghill, G. G. 430] Colin, Andrew, 128] Collard, Philippe, 129] Collins, Robert James, 354, 567, 568] Colombetti, Marco, 168, 169] Colorni, Alberto, 166, 171, 463, 464] Colvin, M. E. 360] ....

[Article contains additional citation context not shown here]

David T. Cliff, Philip Husbands, and Inman Harvey. Analysis of evolved sensory-motor controllers. Technical Report CSRP264, University of Sussex, School of Cognitive and Computing Sciences, 1992. (also as [?]; available via anonymous ftp cite ftp.cogs.susx.ac.uk directory /pub/reports/csrp file csrp264.ps.Z ) ga:Harvey92e.

....the world in simple terms. 1 Introduction In the behavior based approach to robotics, there is a strong interest in exploiting the phenomenon of emergent behavior for the development of advanced robot controllers. One of the most clear cut ways in this direction is evolutionary robotics, cf. [15, 14, 10, 3, 9, 11, 2, 16, 12] which tries to breed behaviors oriented by conveniently defined fitness functions. Another way is reinforcement learning where the behaviors are driven into the desired directions by the rewards obtained from the environment. However despite some interesting achievements, there is also a lot of ....

D. Cliff, P. Husbands, and I. Harvey. Analysis of evolved sensory-motor controllers. Technical Report Cognitive Science Research Paper CSRP264, School of Cognitive and Computing Sciences, University of Sussex, Brighton BN1 9QH, England, UK, 1992.

.... first papers, we advocate [ the concurrent evolution of visual sensor morphology and the control networks: separating mor Funes and Pollack Evolutionary Body Buildnig 9 17 98 3 phology from control is [ difficult to justify from an evolutionary perspective and potentially misleading [7]. But evolution of robot bodies remains limited mostly to adjusting parameters of the sensory configuration (sensor positioning, and direction, receptive fields, and so on) 8, 27] Recent work by Lund, Hallam and Lee [29] addresses some morphological issues, such as body size and wheel radius ....

Cliff, D., Husbands, P.and Harvey, I. (1992) Analysis of Evolved Sensory-Motor Controllers. Technical Report CSRP 264, University of Sussex School of Cognitive and Computing Sciences.

.... intelligent agents are those that have the capacity to flourish by means of suitably adjusting their interactions with an environment on which they depend for resources, even though the agent (and the agent s designer, if any) cannot safely make a priori assumption about the environment [Varela91, Parisi92, Nolfi93, Cliff93a, Cliff93b, Steels94]. Adaptive evolution and kindred processes are especially salient sources of this kind of sensorimotor functionality [Bedau91, Varela91, Parisi92, Nolfi93, Cliff93a, Cliff93b, Steels94] indeed, they may be its only source. Thus, a fundamental research goal is to understand how the process of ....

.... even though the agent (and the agent s designer, if any) cannot safely make a priori assumption about the environment [Varela91, Parisi92, Nolfi93, Cliff93a, Cliff93b, Steels94] Adaptive evolution and kindred processes are especially salient sources of this kind of sensorimotor functionality [Bedau91, Varela91, Parisi92, Nolfi93, Cliff93a, Cliff93b, Steels94] indeed, they may be its only source. Thus, a fundamental research goal is to understand how the process of adaptive evolution creates and shapes sensorimotor functionality. Those effects of action on perception involved in sensorimotor functionality take at least two forms. An agent s actions ....

[Article contains additional citation context not shown here]

D. Cliff, I. Harvey, and P. Husbands. Analysis of evolved sensory-motor controllers. In Proceedings of the Second European Conference on Artificial Life (ECAL93), Brussels, Belgium, 1993.

....in these experiments is the relatively low fitness of the robots even the best robots are unable to clear the arena completely. The primary reason for this shortcoming is that the robots often get stuck in repetitive cyclic paths. This can be handled by introducing noise in the system [4], thereby allowing the robots to break out of such fixed cycles. As some researchers have argued [4, 10] such noisy systems are of particular importance if controller designs evolved in simulation are to be successfully transferred on to real robots. Sensory system design, in our view, has two ....

....unable to clear the arena completely. The primary reason for this shortcoming is that the robots often get stuck in repetitive cyclic paths. This can be handled by introducing noise in the system [4] thereby allowing the robots to break out of such fixed cycles. As some researchers have argued [4, 10], such noisy systems are of particular importance if controller designs evolved in simulation are to be successfully transferred on to real robots. Sensory system design, in our view, has two main applications. First, it allows us to determine efficient, often counterintuitive designs for the ....

D. Cliff, P. Husbands, and I. Harvey. Analysis of evolved sensory-motor controllers. In Second European Conference on Artificial Life, 1993.

....behaviors, but also say that careful simulations of the robot and of the robot environment interactions can be necessary because of time constraints. Developing on these lines, they present results of several evolved behaviors for a simulated robot with a very simple visual system [55] [56], 57] 14] Floreano [35] has studied the evolution of a simulated agent who developed the ability to reach a nest where it could eat the food found in the external environment. Since the fitness function was simply the number of food objects eaten, the location of the nest and the ability to ....

D. Cliff, P. Husbands, and I. Harvey, "Analysis of Evolved Sensory-Motor Controllers", Tech. Rep. CSRP 264, School of Cognitive and Computing Science, University of Sussex, December 1992.

.... published in (Harvey 1993a) Chapter 9 appeared in similar form in (Harvey 1992a) Applications of these ideas to robots in simulation and in reality, discussed in Chapter 12 and 13, have been covered in many jointly written papers: Harvey et al. 1993c) Harvey et al. 1993a) Harvey et al. 1993b) (Cliff et al. 1993g) Cliff et al. 1993a) Cliff et al. 1993d) Cliff et al. 1993c) Cliff et al. 1993b) Cliff et al. 1993e) Cliff et al. 1993h) Cliff et al. 1993f) Husbands et al. 1993a) Husbands et al. 1995) Husbands et al. 1993c) Husbands et al. 1993b) This work was done jointly with Phil Husbands and ....

.... 1993a) Chapter 9 appeared in similar form in (Harvey 1992a) Applications of these ideas to robots in simulation and in reality, discussed in Chapter 12 and 13, have been covered in many jointly written papers: Harvey et al. 1993c) Harvey et al. 1993a) Harvey et al. 1993b) Cliff et al. 1993g) (Cliff et al. 1993a) Cliff et al. 1993d) Cliff et al. 1993c) Cliff et al. 1993b) Cliff et al. 1993e) Cliff et al. 1993h) Cliff et al. 1993f) Husbands et al. 1993a) Husbands et al. 1995) Husbands et al. 1993c) Husbands et al. 1993b) This work was done jointly with Phil Husbands and Dave Cliff. The ....

[Article contains additional citation context not shown here]

D. T. Cliff, P. Husbands, and I. Harvey. Analysis of evolved sensorymotor controllers. In Proceedings of Second European Conference on Artificial Life, ECAL93. Brussels, May 1993, 1993.

....(c.f. input, hidden, and output layers found in back propagation networks) Nevertheless, for the purposes of generating adaptive behaviour, it is necessary to designate some units as receiving input from sensors, and others as producing outputs to actuators (such as motors) As is discussed in [3], this designation may be distorted by the evolutionary processes. The remainder of this section discusses details of the neuron model, and how the networks architectures are encoded as genes which can be operated on by the saga genetic algorithm. 2.1 The Neuron Model The neuron model we have ....

....network was not designed by a human: it evolved according to Darwinian principles. Yet the plots of the activity of the networks clearly indicate that the robot is approaching the centre of the arena and staying there: they achieve their specified task. For further analysis of these networks, see [3]. The next 1 The simulations involve accurate physical and kinematic models of a real robot constructed at Sussex. Vision was simulated using ray tracing with anti aliasing via 16 fold super sampling. See [4] for further details. section discusses our findings from studies of varying the ....

[Article contains additional citation context not shown here]

D. T. Cliff, P. Husbands, and I. Harvey. Analysis of evolved sensory motor controllers. Technical Report CSRP 264, University of Sussex School of Cognitive and Computing Sciences, 1992.

....Typically behaviour interpretable as sensible appeared in less than 100 generations, using a population of size 60; a high scoring trajectory is shown in Fig. 4, and the network that produced this behaviour in Fig. 5. For further analysis of such networks, and how they produce the behaviour, see [3]. The present paper concentrates on the issues of genetic convergence. The evolutionary principles on which these experiments are based allow for incremental adding of tasks, requiring additional new behaviours or changes in old ones. The present analysis, however, is restricted to just a first ....

D. T. Cliff, P. Husbands, and I. Harvey. Analysis of evolved sensory-motor controllers. Technical Report CSRP 264, School of Cognitive and Computing Sciences, Univ. of Sussex, 1993.

....experiments the cylinder radius was fixed at 20 units and the wall height at 15 units. The robot radius was 2 units. Success on this task was reached on a number of runs, within 100 generations each time. Two different successful control systems, termed C1 and C2, have been described elsewhere [11]. In both cases, the robots make a smooth approach towards the center of the arena, and then circle there, either on the spot or in a minimum radius circle. A typical behaviour for C1 is shown in Figure 4 and the network that generated it is illustrated in Figure 5. Compare these with the ....

D. T. Cliff, P. Husbands, and I. Harvey. Analysis of evolved sensory-motor controllers. In Proceedings of Second European Conference on Artificial Life, ECAL93. Brussels, May 1993, 1993.

....of photosensors on the robot s body. The robot was evolved to perform a simple visually guided behavior: to find its way to the center of a circular room. The resulting controller networks were analyzed using a number of techniques: qualitative studies of the functional architecture of the network (Cliff, Husbands Harvey 1992) and the effects of noise (Cliff, Harvey Husbands 1993a) and quantitative studies based on dynamical systems analysis (Husbands, Harvey Cliff 1995) Sims (1994) demonstrated a methodology for jointly evolving morphologies and controllers for embodied three dimensional creatures. The creatures ....

Cliff, D., Husbands, P. & Harvey, I. (1992), Analysis of Evolved Sensory Motor Controllers, Technical Report CSRP 264, School of Cognitive and Computing Sciences, Sussex University. Presented at the Second European Conference on Artificial Life (ECAL93); unpublished proceedings.

No context found.

Cliff, D., Husbands, P. & Harvey, I. "Analysis of evolved sensory-motor controllers." 2nd European Conference on Artificial Life, Brussels. 1992

No context found.

D. Cliff, P. Husbands, I. Harvey, University of Sussex, Analysis of Evolved Sensory-Motor Controllers (1992).

No context found.

Cliff, D., P. Husbands, and I. Harvey (1993a). Analysis of evolved sensory motor controllers.

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