L. Steels. The artificial life roots of artificial intelligence. Artificial Life, 1(1-2):75--110, 1994.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....A behavior is intelligent to the extent that it maximises the agent s chances for self preservation. The unpredictability of physical effects in the agent and in the environment is crucial for requiring the physical realisation and appearance of the agent in its surroundings ( Brooks91] [Steels94]) On the other hand, the ability to sense and act intelligently stands and falls with the design of so called behavior systems which reside inside the agent and define the possibilities of interaction with the environment. There are several different approaches to implement behavior systems: ....

Luc Steels. The artificial life roots of artificial intelligence. Artificial Life Journal, 1,1, 1994. 4

....and discussed. 1998 Elsevier Science B.V. All rights reserved. Keywords: Autonomous robots; Fuzzy; Genetic algorithms; Learning 1. Introduction Artificial Life has been defined as a scientific dis cipline that studies how behavior of agents emerges and becomes intelligent and adaptalive [13]. Many experiments have been made in this sense using neural networks, classifier systems, etc. showing how these behaviors can emerge. This paper presents the evolution of the behavior of a robot using a different paradigm: fuzzy logic. In this way, the evolution of the control rules for an ....

L. Steels, The artificial life roots of artificial intelligence, Artifical Life 1 (1994) 75-110.

....robots are by no means a new idea. During the 1950s a vacuum tube robot turtle built by W. Grey Walter could recharge itself [Walter, 1950] Recharging robots built by Luc Steels have begun to demonstrate the potential of having learning robots cope with internal and external energy management [Steels, 1993] [Steels, 1994] Learning mobile robots that cannot recharge are limited by their short run times. In such cases, the learning must be scaled down to fit into the lifetime of the battery pack. In her 1994 Ph.D. dissertation, Maja Mataric noted the limitations of short run time robots [Mataric, ....

....occurred, the weaker robot was unable to push the stronger one off the charger and would spin its wheels until its sensors finally saw the robot as an obstacle, at which point it would give up. Within several hours, the weaker robot was dead and the stronger robot ran for another day. Luc Steels [Steels, 1993] defines emergent behavior as follows: A behavior is emergent if it can only be defined using descriptive categories which are not necessary to describe the behavior of the constituent components. An emergent behavior leads to emergent functionality if the behavior contributes to the system s ....

Luc Steels, "The Artificial Life Roots of Artificial Intelligence." Artificial Life Journal, Vol. 1,1. MIT Press, Cambridge.

....Machines only consume energy. Thus, the label of artificial life , which suggests that a technical reconstruction of life forms can be done, is, in our opinion, even more unfortunate and prone to misunderstanding than the label of artificial intelligence . Reading overview papers (like [St 93] one gets in fact the impression that the field is at least in part a jumble of incoherent and ill defined philosophical, biological and computer science based positions with questionable (neo frankenstein like) aspirations 4 . Our goals are theory oriented and can be simply viewed in the ....

Steels, L., The artificial life roots of artificial intelligence, AI Lab Univ. Brussel, AI-Memo 93-80

....easily have marked a recharging point but in these experiments only triggered off behaviour leading to the energy level being restored to full. A similar scenario has often been used in research studying survival and exploration of physical mobile robots in an artificial life experimental set up [11, 12]. Experiments are conducted in environments of differing size. Complexity is increased by adding more wall structures (as shown in figure 2) In this way the applicability of the robot to varying environments was tested. Also, experiments were carried out in environments containing other robots ....

L. Steels, (1994), The artificial life roots of artificial intelligence, Artificial Life Journal 1(1), MIT Press, Cambridge, MA, pp. 89-125.

....abstract problems (e.g. scheduling work plans, optimizing parameters, etc. it will be cumbersome to create a detailed model of a noisy environment. Modeling all the small perturbations and small details of the real world is next to impossible. Therefore behavior oriented robotics [5] 43] [68] has taken an all together di#erent approach. It tries to use the world as its own best model , i.e. a symbolic modeling of the world is avoided by constantly monitoring the real world. This is a significant shift of the viewpoint. Instead of viewing the infinite number of real world details as ....

....the robot building box. It allows fast changes of the robot configuration to test di#erent robot architectures. In particular M. Asada s group has a long history of investigating machine learning by using soccer experiments [2] Investigation of robots in other scenarios includes the VUB robots [68] [47] the robots at the University of Bielefeld [63] or the Nerd Herd at Brandeis University [44] The most important development in modern design automation is codesign of hardware and software. One of the most prominent examples in this context is Ptolemy [8] It aims at integrating ....

L. Steels. The artificial life roots of artificial intelligence. Artificial Life Journal, 1(1), 1994.

....is no absolute security as trusted systems can cheat. But the process is completely open and robust as trust is not predefined, but it emerges from subtle interactions between the systems. The basic ideas of this process go back to two roots, namely the field of Artificial Life or short Alife [Ste94a, Lea90, Lan89] and the field of Evolutionary Game Theory [Smi84, Axe84, AH81, SP73] Before the process of the formation of trust can be described, it is necessary to first define the notion of trust itself as it is used here. The basis for trust is an intrinsic property of each individuum in form of the ....

Luc Steels. The artificial life roots of artificial intelligence. Artificial Life Journal, Vol 1,1, 1994.

....and sequential patterns of activity in multiple output systems. Is There a Need for Specialized Selection Circuitry in Complex Control Architectures Emergent action selection. Work in the field of adaptive behavior has been at the forefront of the study of emergent functionality (see, e.g. Steels, 1995) where useful behavioral outcomes are seen to arise as a consequence or side effect of the interaction of control system components which individually have a different or more limited functionality. In such systems no one component is decisive in shaping the overall outcome, and it is generally ....

Steels, L. (1995). The artificial life roots of artificial intelligence. In C. G. Langton (Ed.), Artificial life: an overview (pp. 75--110). Cambridge, MA: MIT Press.

....from local interactions. Thermodynamic emergence is concerned with issues such as the origins of life, where order emerges from noise. The emergence relative to a model concept deals with situations where observers need to change their model to keep up with a system s behaviour. This is close to Steels (1994) concept of emergence, which refers to ongoing processes which produce results invoking vocabulary not previously involved in the description of the system s inner components new descriptive categories (section 4.1) Evolutionary emergence falls into the emergence relative to a model ....

Steels, L. (1994). The artificial life roots of artificial intelligence. Artificial Life Journal 1(1), 89--125.

.... the functionally based view when he writes: Emergent functionality means that a function is not achieved directly by a component or a hierarchical system of components, but indirectly by the interaction of more primitive components among themselves and with the world [my italics] Later (Steels 1994), he refers to ongoing processes which produce results invoking vocabulary not previously involved in the description of the system s inner components new descriptive categories . The with the world qualification is potentially important and often forgotten (although not by Gell Mann 14 or ....

Steels, L. (1994). The artificial life roots of artificial intelligence. Artificial Life Journal 1(1), 89--125.

....not at least for communication purposes construct its own symbols, but grounds the user defined symbols onto its own perceptions and actions. This point of view is clearly in contrast to approaches in artificial life (AL) Stewart, 1995) Complete autonomy the main goal of research in AL (Steels, 1994) and the emphasis on learning for survival (i.e. to enable the robot to wander around for several days without colliding with obstacles and to learn when it is necessary to re charge the battery (Birk, 1996) is less important here. In these approaches, the systems learn from the environment, ....

Steels, L. (1994). The artificial life roots of artificial intelligence. Artificial Life, 1(1).

....adaptive control rule based incremental control fuzzy control neural control adaptive fuzzy control A few comments are necessary, mostly concerning fuzzy and neural networks applied to control. Indeed many different neural networks and fuzzy architectures have been proposed. As discussed in [Ste94] an advantage of neural network approaches is that they immediately incorporate a mechanism for learning. However, a major disadvantage is that the global search space is too big to start from zero, and much initial structure must typically be encoded which is sometimes difficult to express in ....

L. Steels. The artificial life roots of artificial intelligence. Artificial Life Journal, 1(1), 1994.

....its import for embodied systems. Embodied Systems An embodied system takes the physical world as its external environment. 1 Researchers working on evolutionary robotics[6] are exploring how evolutionary mechanisms can be used to create control architecture for robots. Also, in some circles[1][11], much emphasis is placed on the nature of the interaction between a robot s sensors and its actuators for informing the types of behaviours that will be produced. However, from Cariani s formulation of emergence, we will only be able to attain an openly evolving, embodied system if we are able ....

Steels, L. "The Artificial Life Roots of Artificial Intelligence", in Artificial Life III, Langton, C. G., ed. Addison--Wesley Publishing Co., Redwood City, CA, 1994.

....to change their model in order to keep up with a system s behavior. This is close to Steels concept of emergence, which refers to ongoing processes which produce results invoking vocabulary not previously involved in the description of the system s inner components new descriptive categories (Steels, 1994, section 4.1) Evolutionary emergence falls into the emergence relative to a model category. Consider a virtual world of organisms that can move, reproduce and kill according to rules sensitive to the presence of other organisms, evolving under natural selection. Should flocking manifest ....

Steels, L. (1994). The artificial life roots of artificial intelligence.

.... systems take inspiration from biology, ethology, and neuroscience, in order to construct controllers for agents situated in noisy and dynamic environments (Matari c 1997) Behaviors impose a distributed, bottom up approach to control, which eliminates bottlenecks and enables emergent functionality (Steels 1994). Aside from its pragmatic role in achieving distributed, real time control in robotics, the behaviorbased framework has also been used to model and implement simulations of biological systems. In this paper we apply behavior based control to a novel domain, that of biologically inspired 3D ....

Steels, L. (1994), `The Artificial Life Roots of Artificial Intelligence ', Artificial Life 1(1), 75--110.

....(1995) for a delightful collection of essays on android epistemology. 2. This is perhaps an overstatement, since researchers with strong roots in artificial life (alife) and robotics traditions have continued to make significant contributions to our understanding of autonomous agents (Maes 1993; Steels 1995). Although most researchers in robotics have concerned themselves with agents embodied in hardware, some have also made significant contributions in the area of software agents. See Etzioni (1993) for arguments that software presents a no less attractive platform than hardware for the ....

Steels, L. 1995. The Artificial Life Roots of Artificial Intelligence. In Artificial Life: An Overview, ed. C. G. Langton, 75--110. Cambridge, Mass.: MIT Press.

No context found.

Steels, L. (1994) The Artificial Life Roots of Artificial Intelligence. Artificial Life Journal 1(1), pp. 89-125.

....attention to the topic. Their perceptual capabilities must be the basis of the discrimination games and finally they must realise the language games themselves. The robots used in the experiments are Lego vehicles built for our laboratory s experiments in self sufficient robots (see figure 1) [6]. Each robot (size: 30 x 20 x 15 cm) has three infra red sensors (mounted on the leftfront, middle front and right front side) four infrared emitters (mounted on front, left, right, and back side) two visible light sensors (mounted on left and right front side) two modulated light sensors ....

....MC86332 micro controller with 128 kB ROM and 256 kB RAM located on a Vesta board. Its CPU is 16.78 MHz at 5V. The Vesta board is extended with a second board dedicated to low level sensory motor processing and buffering [14] The robots are programmed using a behavior oriented architecture [6]. The sensors, actuators and internal states constitute continuous data streams and the behavior is based on continous dynamical systems implementing direct couplings between sensors and actuators. An 3 0 50 100 150 200 250 1 15 29 43 57 71 85 99 113 127 141 155 169 183 197 211 225 239 253 ....

Steels, L. (1994) The Artificial Life Roots of Artificial Intelligence. 1(1), pp. 89-125.

....visible light, sound, touch, etc. actuators for moving around in the environment, batteries, and on board processors. The robots operate in a physical ecosystem in which they have opportunities to recharge their batteries but also competitors which have to be countered by performing work [4]. Experiments are going on to carry to port the mechanisms reported in this paper onto the physical robots operating in this ecosystem (see [7] for a more extended discussion) The various mechanisms described in the present paper are mapped on a physical counterpart as follows. The sensory ....

Steels, L. (1994) The Artificial Life Roots of Artificial Intelligence. Artificial Life Journal 1(1), pp. 89-125.

No context found.

L. Steels. The artificial life roots of artificial intelligence. Artificial Life, 1(1-2):75--110, 1994.

No context found.

L. Steels. The artificial life roots of Artificial Intelligence. Artificial Life Journal, 1(1):89--125, 1994.

No context found.

L. Steels. The artificial life roots of artificial intelligence. Artificial Life, 1(1):1--86, 1994.

No context found.

L. Steels. The artificial life roots of artificial intelligence. Artificial Life Journal, 1(1):89--125, 1994.

No context found.

Steels, Luc. "The Artificial Life Roots of Artificial Intelligence," in C. Langton (ed.) Artificial Life IV. Addison-Wesley, 1994, 75-110.

No context found.

Steels, L.: 1994a, `The artificial life roots of artificial intelligence'. Artificial Life Journal, Vol 1,1.

First 50 documents  Next 50

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