Explorations in Evolutionary Robotics

By D. Cliff, P. Husbands, Inman R. Harvey. 1993

We discuss the methodological foundations for our work on the development of cognitive architectures, or control systems, for situated autonomous agents. Our focus is the problems of developing sensorimotor control systems for mobile robots, but we also discuss the applicability of our approach to the study of biological systems. We argue that, for agents required to exhibit sophisticated interactions with their environments, complex sensorimotor processing is necessary, and the design, by hand, of control systems capable of such processing is likely to become prohibitively difficult as complexity increases. We propose an automatic design process involving artificial evolution, wherein the basic building blocks used for evolving cognitive architectures are noise-tolerant dynamical neural networks. These networks may be recurrent and should operate in real time. The evolution should be incremental, using an extended and modified version of a genetic algorithm. Practical constraints suggest that initial architecture evaluations should be done largely in simulation. To support our claims and proposals, we summarize results from some preliminary simulation experiments in which visually guided robots are evolved to operate in simple environments. Significantly, our results demonstrate that robust visually guided control systems evolve from evaluation functions that do not explicitly require monitoring visual input. We outline the difficulties involved in continuing with simulations and conclude by describing specialized visuorobotic equipment, designed to eliminate the need for simulated sensors and actuators.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Born to Learn: the Inspiration, Progress, and Future of Evolved Plastic Artificial Neural Networks (2017, background): “Evolutionary robotics (ER) (Cliff et al., 1993; Floreano and Mondada, 1994, 1996; Urzelai and Floreano, 2000; Floreano and Nolfi, 2004) contributed particularly to the development of EPANNs, providing a testbed for applied controllers in robotics.”
Creativity and Artificial Intelligence (1998, background): “Likewise, certain jobs in the field of evolutionary robotics have created new sensory anatomy and control systems created by GAS that can modify the length of the “genome” [4].”
Quality Diversity: A New Frontier for Evolutionary Computation (2016, background): “The products of nature have long served as inspiration for the investigation and practice of evolutionary algorithms and evolutionary robotics (Cliff et al., 1993; Nolfi and Floreano, 2000; Stanley, 2011).”
The evolvability of artificial neural net-works for robot control (2002, methodology): “The field of evolutionary robotics applies the techniques of evolutionary computation to the design of robot controllers (Husbands and Harvey, 1992; Brooks, 1992; Cliff et al., 1993; Floreano and Mondada, 1994; Husbands and Meyer, 1998; Nolfi and Floreano, 2000).”
Competitive Coevolution through Evolutionary Complexification (2011, methodology): “Expanding the length of the size of the genome has been found eective in previous work ( Cli, Harvey, & Husbands, 1993; Harvey, 1993; Koza, 1995; Lindgren & Johansson, 2001).”
How to Evolve Autonomous Robots: Different Approaches in Evolutionary Robotics (1994, background): “In particular we should answer the questions: What to evolve? And, how to evolve it? In the last few years new approaches that involve a form of simulated evolution have been proposed in order to build autonomous robots that can perform useful tasks in unstructured environments (Brooks, 1992; Cliff”
The evolved radio and its implications for modelling the evolution of novel sensors (2002, background): “For example, different researchers have c oevolved robot sensor morphology and controllers [ 3 ,4] which can lead to insights into how particular environmental conditions affect sensory processing by agents [5].”
Evolution of Plastic Control Networks (2001, methodology): “DF thanks the Jet Propulsion Laboratory for the invitation to the workshop on Biomorphic Robotics and the referees for useful comments that helped to improve this article.”
Embodied Evolution: Distributing an evolutionary algorithm in a population of robots (2002, background): “Our most intriguing, yet currently unexplored, motivation for EE arises from the fact that, since EE uses a distributed EA in a population of real robots, it may provide an effective way to investigate interactive and collective tasks.”
The Dynamics of Active Categorical Perception in an Evolved Model Agent (2003, methodology): “In the work described here, evolutionary algorithms (Mitchell, 1996) are used to evolve dynamical nervous systems for situated model agents (Beer & Gallagher, 1992; Cliff , Harvey, & Husband, 1993; Nolfi & Floreano, 2000).”
Artificial Evolution in the Physical World (1997, methodology): “When evolving control systems for autonomous mobile robots [1], even when the real robot is used instead of a simulation, the actual structure undergoing evolution | often an arti cial neural network (ANN) | is usually simulated in software.”
Evolving Mobile Robots in Simulated and Real Environments (1995, methodology): “Although within the Evolutionary Robotics community there are examples of controllers for autonomous robots implemented with explicit programs written in a high-level language [2, 12] or with a form of classifier systems [6], many researchers decided to use neural networks [4, 7, 14, 17, 21].”