We address two issues in Evolutionary Robotics, namely the genetic encoding and the performance criterion, also known as fitness function. For the first aspect, we suggest to encode mechanisms for parameter self-organization, instead of the parameters themselves as in conventional approaches. We argue that the suggested encoding generates systems that can solve more complex tasks and are more robust to unpredictable sources of change. We support our arguments with a set of experiments on evolutionary neural controller for physical robots and compare them to conventional encoding. In addition, we show that when also the genetic encoding is left free to evolve, artificial evolution will select to exploit mechanisms of self-organization. For the second aspect, we shall discuss the role of the performance criterion, also known as fitness function, and suggest Fitness Space as a framework to conceive fitness functions in Evolutionary Robotics. Fitness Space can be used as a guide to design fitness functions as well as to compare different experiments in Evolutionary Robotics. 1 1

This paper is concerned with adaptation capabilities of evolved neural controllers. We propose to evolve mechanisms for parameter self-organization instead of evolving the parameters themselves. The method consists of encoding a set of local adaptation rules that synapses follow while the robot freely moves in the environment. In the experiments presented here, the performance of the robot is measured in environments that are different in significant ways from those used during evolution. The results show that evolutionary adaptive controllers solve the task much faster and better than evolutionary standard fixed-weight controllers, that the method scales up well to large architectures, and that evolutionary adaptive controllers can adapt to environmental changes that involve new sensory characteristics (including transfer from simulation to reality and across different robotic platforms) and new spatial relationships.

For many years, researchers in the field of mobile robotics have been investigating the use of genetic and evolutionary computation (GEC) to aid the development of mobile robot controllers. Alongside the fundamental choices of the GEC mechanism and its operators, which apply to both simulated and physical evolutionary robotics, other issues have emerged which are specific to the application of GEC to physical mobile robotics. This paper presents a survey of recent methods in GEC-developed mobile robot controllers, focusing on those methods that include a physical robot at some point in the learning loop. It simultaneously relates each of these methods to a framework of two orthogonal issues: the use of a simulated and/or a physical robot, and the use of finite, training phase evolution prior to a task and/or lifelong adaptation by evolution during a task. A list of evaluation criteria are presented and each of the surveyed methods are compared to them. Analyses of the framework and evaluation criteria suggest several possibilities; however, there appear to be particular advantages in combining simulated, training phase evolution (TPE) with lifelong adaptation by evolution (LAE) on a physical robot.

After reviewing current approaches in Evolutionary Robotics, we point to directions of research that are likely to bring interesting results in the future. We then address two crucial aspects for future developments of Evolutionary Robotics: choice of fitness functions and scalability to real-world situations. In the first case we suggest a framework to describe fitness functions, choose them according to the situation constraints, and compare available experiments in the literature on evolutionary robotics. In the second case, we suggest a way to make experimental results applicable to realworld situations by evolving online continuous adaptive controllers. We also give an overview of recent experimental results showing that the suggested approaches produce qualitatively superior abilities, scale up to more complex architectures, smoothly transfer from simulations to real robots and across different robotic platforms, and autonomously adapt in few seconds to several sources of strong variability that were not included during the evolutionary run. 1

Reports on evolutionary robotics systems have so far been on evolving controllers that make simple robots do simple tasks in simple environments. In this paper we try to stress the evolutionary robotics approach by evolving a controller for a more complex task, namely Khepera robot soccer, and evaluate evolved controller performance against handcoded controllers. We present a system that uses competitive co-evolution to develop robot controllers for the task. The system is described, and performance of the system is documented. Co-evolution is tested against single-population evolution, and it is concluded that co-evolution has the ability to produce more robust individuals with respect to opponent strategies.

The paper introduces homeokinesis as a general principle for the self-organization of robot behavior. Homeokinesis is a domaininvariant principle which induces specific, seemingly goal--oriented behaviors of an agent in a complex external world. The principle is based on the assumption that the agent is able of (i) learning an internal representation (self-model) of its current behavior and (ii) to adapt its behavior in such a way that the complexity gap between model and true behavior is minimized. Driven by the basic impetus of always moving and acting the agent in a noisy complex world develops smooth controlled behaviors as the minimal compromise between the high dynamical complexity of the world and the low complexity of the model. In these terms behavior may be called a byproduct (epiphenomenon) of the internal "desire" of the agent to be able of interpreting the world in simple terms. 1 Introduction Can the self-organization of behavior from a practical point of v...

We review recent experiments in evolutionary robotics carried out in dynamic environments and across different robotic platforms. We then introduce a new evolutionary approach where robots are evolved for their ability to adapt online. Several experiments show that this new approach is much faster, more powerful, and scalable than the traditional approach. 1 Evolutionary Robotics Autonomous robots are largely replacing computers as a metaphor for investigating natural and artificial intelligent systems because they interact with a real environment through sensors and actuators in a closed feedback loop, are subject to the laws of physics, operate in real-time, and are required to cope with partially unknown and unpredictable situations. Artificial evogenetic. method.eps 66 \Theta 41 mm ... Mutation Crossover Selective reproduction Evaluation Population manager Figure 1: A single physical robot is connected to a host computer through a serial cable with rotating contacts. The ...

Abstract. We provide an overview of the evolutionary approach to the emergence of artificial intelligence in embodied behavioral agents. This approach, also known as Evolutionary Robotics, builds and capitalizes upon the interactions between the embodied agent and its environment. Although we cover research carried out in several laboratories around the world, the choice of topics and approaches is based on work carried out at EPFL. We describe a large number of experiments including evolution of single robots in environments of increasing complexity, competitive and cooperative evolution, evolution of vision-based systems, evolution of learning, and evolution of electronics and morphologies for autonomous robots. 1

Evolutionary Design has been used to automatically generate a wide variety of novel and creative objects such as circuits, robots, and satellite antennae. And yet, despite the availability of sophisticated rapid prototyping machines capable of printing objects out of plastic, metal, and even circuitry, relatively few of these evolved designs have been physically manufactured in the real world. We argue

This paper reviews the use of artificial evolution as a means for automatic generation of locomotion controllers for physical robots and autonomous agents from the perspectives of evolutionary objectives and control architecture. An overview beginning with the pioneering works in evolutionary robotics is given, leading up to the latest state-of-the-art research in these fields. Some key shortfalls in mainstream approaches are identified, concluding with some promising research directions. Keywords: Evolutionary robotics, evolutionary artificial neural networks, fitness functions, controller architecture, autonomous robots.