Evolving Control Metabolisms for a Robot

Abstract

This article demonstrates a new method of programming artificial chemistries. It uses the emerging capabilities of the system's dynamics for information-processing purposes. By evolution of metabolisms that act as control programs for a small robot one achieves the adaptation of the internal metabolic pathways as well as the selection of the most relevant available exteroceptors. The underlying artificial chemistry evolves efficient information-processing pathways with most benefit for the desired task, robot navigation. The results show certain relations to such biological systems as motile bacteria.

Controlling Complex Dynamics with Artificial Biochemical Networks

“…An artificial chemistry consists of a set of chemicals, a set of rules -which model the transformative agents (such as enzymes) found in natural systems -and an algorithm that determines when these rules are applied. Chemicals may be symbols to which some computational meaning can be associated [19], they may directly encode data structures, they may be overtly computational in nature, e.g. lambda-expressions [10], or they may even be other ABNs [8].…”

“…Likewise, rules vary from simple symbolic transformations to functional composition and complex structural modifications. By encoding inputs and outputs in the concentration, internal structure or positioning of chemicals, these artificial chemistries have been applied to a number of computational tasks, including robotics [19].…”

“…This has led to growing interest in the use of computational models of biochemical networks, particularly within genetic programming. These include artificial genetic regulatory networks [2,12,17,18], computational models of cellular metabolism [1,8,10,19], and those derived from signalling networks [6]. We refer to them collectively as artificial biochemical networks (ABNs).…”

“…Traditionally, robotic control has been a popular application area for ABNs [7,17,19]. In principle, it reflects one of the main biological roles of biochemical networks: maintaining correct behaviour when exposed to a complex, dynamic, environment.…”

An Algorithmic Chemistry for Genetic Programming

“…Following previous work on Artificial Chemistries (see, for example [5,6,7,8]), [9] introduces a very general analogy between chemical reaction and algorithmic computation, arguing that concentrations of results would be important. [10] was the first step in this new direction.…”