Abstract not found

This paper puts forward a theoretical framework and formal language for understanding how simple, fast simulations can be used to artificially evolve controllers for real robots. It begins by putting forward a general set of equations that describe the way in which an agent-environment system changes over time, and analyses what it means for an agent to exhibit a particular behaviour within such a system. A minimally sufficient set of conditions are then formally established under which a controller that reliably displays a particular behaviour in one system will continue to display the same behaviour when transplanted into another system, and the special case of behavioural transference from simulation into reality is evaluated within the context of this theoretical framework. From this, techniques are derived for ensuring that controllers which are reliably fit within a simulation will transfer into reality, and two sets of experiments are briefly described in which controllers that ...

We present a model of decentralized growth and development for artificial neural networks (ANNs), inspired by developmental biology and the physiology of nervous systems. In this model, each individual artificial neuron is an autonomous unit whose behavior is determined only by the genetic information it harbors and local concentrations of substrates. The chemicals and substrates, in turn, are modeled by a simple artificial chemistry. While the system is designed to allow for the evolution of complex networks, we demonstrate the power of the artificial chemistry by analyzing engineered (handwritten) genomes that lead to the growth of simple networks with behaviors known from physiology. To evolve more complex structures, a Java-based, platform-independent, asynchronous, distributed genetic algorithm (GA) has been implemented that allows users to participate in evolutionary experiments via the World Wide Web.

On behalf of:

This article describes past and current research efforts in evolutionary robotics that have been carried out at the AnimatLab, Paris. Such approaches entail using an artificial selection process to automatically generate developmental programs for neural networks that control rolling, walking, swimming and flying animats or robots. Basically, they complement the underlying evolutionary process with a developmental procedure – in order hopefully to reduce the size of the genotypic space that is explored – and they occasionally call on an incremental approach, in order to capitalize upon solutions to simpler problems so as to devise solutions to more complex problems. This article successively outlines the historical background of our research, the evolutionary paradigm on which it relies, and the various results obtained so far. It also discusses the potentialities and limitations of the approach and indicates directions for future work.

An incremental approach is used to simulate the evolution of neural controllers for robust obstacle-avoidance in a Khepera robot and proves to be more ecient than a direct approach. During a rst evolutionary stage, obstacle-avoidance controllers in medium-light conditions are generated. During a second evolutionary stage, controllers avoiding stronglylighted regions, where the previously acquired obstacle-avoidance capacities would be impaired, are obtained. The best controllers thus evolved are successfully downloaded on a Khepera robot. The SGOCE paradigm that is used in these experiments is described in the text. Future research will target at furthering the incremental evolutionary process and evolving more intricate behaviors. 1 Introduction According to a recent review [26] of evolutionary approaches to neural control in mobile robots, it appears that the corresponding research eorts usually call upon a direct encoding scheme, where the phenotype of a given robot | ...

This paper gives an overview of the bottom-up approach to artificial intelligence (AI), commonly referred to as behavior-oriented AI. The behavior-oriented approach, with its focus on the interaction between autonomous agents and their environments, is introduced by contrasting it with the traditional approach of knowledge-based AI. Different notions of autonomy are discussed, and key problems of generating adaptive and complex behavior are identified. A number of techniques for the generation of behavior are introduced and evaluated regarding their potential for realizing different aspects of autonomy as well as adaptivity and complexity of behavior. It is concluded that in order to realize truly autonomous and intelligent agents, the behavior-oriented approach will have to focus even more on life-like qualities in both agents and environments.

We present a model of decentralized growth for Artificial Neural Networks (ANNs) inspired by the development and the physiology of real nervous systems. In this model, each individual artificial neuron is an autonomous unit whose behavior is determined only by the genetic information it harbors and local concentrations of substrates modeled by a simple artificial chemistry. Gene expression is manifested as axon and dendrite growth, cell division and differentiation, substrate production and cell stimulation. We demonstrate the model’s power with a hand-written genome that leads to the growth of a simple network which performs classical conditioning. To evolve more complex structures, we implemented a platform-independent, asynchronous, distributed Genetic Algorithm (GA) that allows users to participate in evolutionary experiments via the World Wide Web. 1

Today's `autonomous' robots only have very limited autonomy and are in fact very much under the control of the `environmental puppeteer', i.e their behaviour is determined, via virtual strings, by environmental conditions. Hence, it has been stated as the goal of modern scientific robotics to "cut the strings and give the robot its autonomy". Different notions of autonomy in artefacts and living systems are examined in this paper, and different aspects/dimensions of autonomy are identified and illustrated with examples from connectionist robot control. A connectionist architecture is introduced that aims to increase robotic autonomy through integration of connectionist self-organisation/learning with the enactive view of structural coupling between environment and agent. In the resulting robot control architecture it is the environment that is pulling the strings, but the agent that develops them and dynamically decides which of them to use in a particular situation. Hence...

This paper presents results generated with a new evolutionary robotics (ER) simulation environment and its complementary real mobile robot colony research test-bed. Neural controllers producing mobile robot maze searching and exploration behaviors using binary tactile sensors as inputs were evolved in a simulated environment and subsequently transferred to and tested on real robots in a physical environment. There has been a considerable amount of proof-of-concept and demonstration research done in the field of ER control in recent years, most of which has focused on elementary behaviors such as object avoidance and homing. Artificial neural networks (ANN) are the most commonly used evolvable controller paradigm found in current ER literature. Much of the research reported to date has been restricted to the implementation of very simple behaviors using small ANN controllers. In order to move beyond the proof-of-concept stage our ER research was designed to train larger more complicated ANN controllers, and to implement those controllers on real robots quickly and efficiently. To achieve this a physical robot test-bed that includes a colony of eight real robots with advanced computing and communication abilities was designed and built. The real robot platform has been coupled to a simulation environment that facilitates the direct wireless transfer of evolved neural controllers from simulation to real robots (and vice versa). We believe that it is the simultaneous development of ER computing systems in both the simulated and the physical worlds that will produce advances in mobile robot colony research. Our simulation and training environment development focuses on the definition and training of our new class of ANNs, networks that include multiple hidden layers, and tim...