In this paper we propose and justify a methodology for the development of the control systems, or `cognitive architectures', of autonomous mobile robots. We argue that the design by hand of such control systems becomes prohibitively difficult as complexity increases. We discuss an alternative approach, involving artificial evolution, where the basic building blocks for cognitive architectures are adaptive noise-tolerant dynamical neural networks, rather than programs. These networks may be recurrent, and should operate in real time. Evolution should be incremental, using an extended and modified version of genetic algorithms. We nally propose that, sooner rather than later, visual processing will be required in order for robots to engage in non-trivial navigation behaviours. Time constraints suggest that initial architecture evaluations should be largely done in simulation. The pitfalls of simulations compared with reality are discussed, together with the importance of incorporating noise. To support our claims and proposals, we present results from some preliminary experiments where robots which roam office-like environments are evolved.

One category of researchers in artificial life is concerned with modeling and building so-called adaptive autonomous agents. Autonomous agents are systems that inhabit a dynamic, unpredictable environment in which they try to satisfy a set of time-dependent goals or motivations. Agents are said to be adaptive if they improve their competence at dealing with these goals based on experience. Autonomous agents constitute a new approach to the study of artificial intelligence (AI) which is highly inspired by biology, in particular ethology, the study of animal behavior. Research in autonomous agents has brought about a new wave of excitement into the field of AI. This paper reflects on the state of the art of this new approach.

... the last 5 years. We explain and justify our distinctive approaches to (artificial) evolution, and to the nature of robot control systems that are evolved. Results are presented from research with evolved controllers for autonomous mobile robots; simulated robots, coevolved animats, real robots with software controllers, and a real robot with a controller directly evolved in hardware.

In this report, I discuss the use of vision to support concrete, everyday activity. I will argue that a variety of interesting tasks can be solved using simple and inexpensive vision systems. I will provide a number of working examples in the form of a state-of-the-art mobile robot, Polly, which uses vision to give primitive tours of the seventh floor of the MIT AI Laboratory. By current standards, the robot has a broad behavioral repertoire and is both simple and inexpensive (the complete robot was built for less than $20,000 using commercial board-level components). The approach I will use will be to treat the structure of the agent's activity--- its task and environment---as positive resources for the vision system designer. By performing a careful analysis of task and environment, the designer can determine a broad space of mechanisms which can perform the desired activity. My principal thesis is that for a broad range of activities, the space of applicable mechanisms will be broad...

This paper will look at an evolutionary approach to robotics; partly at pragmatic issues, but primarily at theoretical issues associated with the evolutionary algorithms which are appropriate. Genetic Algorithms are not suitable in their usual form for the evolution of cognitive structures, which must be in an incremental fashion. SAGA -- Species Adaptation Genetic Algorithms -- is a conceptual framework for extending GAs to variable length genotypes, where evolution allows a species of individuals to evolve from simple to more complex. In the context of species evolution the metaphor of hill-crawling as opposed to hill-climbing is introduced,

This paper is concerned with training an agent to perform sequential behavior. In previous work we have been applying reinforcement learning techniques to control a reactive robot. Obviously, a pure reactive system is limited in the kind of interactions it can learn. In particular, it can only learn what we call pseudo-sequences, that is sequences of actions in which the transition signal is generated by the appearance of a sensorial stimulus. We discuss the difference between pseudo-sequences and proper sequences, and the implication that these differences have on training procedures. A result of our research is that, in case of proper sequences, for learning to be successful the agent must have some kind of memory; moreover it is often necessary to let the trainer and the learner communicate. We study therefore the influence of communication on the learning process. First we consider trainer-to-learner communication introducing the concept of reinforcement sensor, which let the learning robot explicitly know whether the last reinforcement was a reward or a punishment; we also show how the use of this sensor induces the creation of a set of error recovery rules. Then we introduce learner-to-trainer communication, which is used to disambiguate indeterminate training situations, that is situations in which observation alone of the learner behavior does not provide the trainer with enough information to decide if the learner is performing a right or a wrong move. All the design choices we make are discussed and compared by means of experiments in a simulated world.

This paper presents an analysis of an artificially evolved dynamical network-based control system for a simulated autonomous mobile robot engaged in simple visually guided tasks.

Designers often improve the performance of artifical agents by specializing them. We can make a rough, but useful distinction between specialization to a task and specialization to an environment. Specialization to an environment can be difficult to understand: it may be unclear on what properties of the environment the agent depends, or in what manner it depends on each individual property. In this paper, I discuss a method for analyzing specialization into a series of conditional optimizations: formal transformations which, given some constraint on the environment, map mechanisms to more efficient mechanisms with equivalent behavior. I apply the technique to the analysis of the vision and control systems of a working robot system in dayto day use in our laboratory.

This paper provides a high-level review of current and recent work in the use of genetic algorithm based techniques to develop sensorimotor control systems for autonomous agents. It focuses on network-based controllers and genetic encoding issues associated with them. The paper closes with a discussion of the possibility of using arti cial evolutionary techniques to help tackle more specifically scientific questions about natural sensorimotor systems.

Abstract not found