Self-replicating Robotic Systems

The concept of an artificial self-replicating system was introduced in the 1950s by John von Neumann [31]. Von Neumann introduced the theory of self-replicating automata and established a quantitativedefinition of self-replication. His early results on self-replicating machines have become useful in several diverse research areas such as:cellular automata, nanotechnology, macromolecular chemistry, and computing [8,23,26,27]. However,prior to the turn of the millennium, a fully autonomous self-replicating physical robot had never been implemented. In this chapter,a series of prototype designs from our laboratory and their physical implementation are described. We begin by discussing some motivation andhistory, then go on to describe a remote-controlled replicating robotic system and a semi-autonomous replicating robotic system. Wethen describe some fully autonomous self-replicating systems, and discuss how manufacturing work cells might be designed so as to reproduce.

Motivation...

Artificial life:

Artificial life refers to the study of artificial systems that exhibit lifelike properties, typically involving some form of self-replication and/or evolution. This is a very broad field of study and includes work involving computer simulations, cellular automata, chemistry, robotics, and synthetic biology.

Cellular automata:

A cellular automaton (CA) is a theoretical construct where a collection of cells are organized into regular grids or lattices. Many arrangements are possible, but typically one-dimensional CAs are composed of square cells arranged in a line, and two-dimensional CAs are composed of square cells arranged in a square grid. Each cell contains a finite state machine. The state of all cells in a CA are typically updated synchronously (at the same time) with each cell changing to a new state that is a function of its previous state and the previous states of its neighbors. The surrounding cells that affect a given cell's state transition are called the cell's neighborhood. CAs can simulate a wide variety of physical processes by designing an appropriate neighborhood and state transition function. CAs are most often studied by implementing them with computer simulations.

Finite state machine:

A finite state machine (FSM) is a conceptual computing machine with an input, output, and memory. At regular intervals in time the machine transitions (changes) to a new state, which is a function of the current state and the machine's current input. The output is a function of the input and the state. The function of the memory is to store the state information between each transition. FSMs are easily implemented with discrete digital electronic components, microcontrollers, or computer simulations. Implementations of FSMs are widely used to control industrial devices and consumer electronics.

Modular robot:

A modular robot is composed of distict “modules” which contain motors, mechanisms, electronics, and interconnections. Modules are typically designed so that they can easily be connected and disconnected from each other. In some modular robotic systems, many identical modules are assembled into larger robots. This allows the same set of modules to form robots optimized for different tasks. In other cases, the modules may be specialized for certain tasks, and when assembled they form a robot with additional functionality. Motivation for building modular robots includes increased versatility, ease of replacing damaged components, and potentially lower manufacturing costs.

Self-reconfigurable robot:

A self-replicating robotic system is a robotic device that exhibits some form of self-replication. This chapter is concerned primarily with directed robotic self-replication, in which a robotic device interprets some form of coded instructions in order to carry out the replication process. In nature deoxyribonucleic acid (DNA) typically encodes replication instructions. In the examples of robotic self-replication presented in this chapter, the instructions may be encoded in a computer program, an arrangement of modular components, or as a pattern of lines that guide the motion of a mobile robot.

Self-replication:

Self-replication is the process by which an entity creates a duplicate of itself. The most familiar example is the self-replication of living organisms, although other natural processes such as crystal growth can be classified as self-replication.

Universal constructor:

A universal constructor (UC) is a conceptual machine that reads instructions and executes them to construct an object. A key property of a UC is that it can construct any object which can be described to it via the instructions, including duplicates of itself. UCs have been demonstrated in computer simulations using cellular automata. Modern manufacturing tools, such as assembly robots and computer-controlled machine tools, are similar to UCs but practical limitations make these machines “somewhat less than universal” constructors. The ribosome, a complex molecule present in nearly all biological cells, performs a function very similar to that of a UC, assembling proteins according to instructions encoded in messenger ribonucleic acid (mRNA).

von Neumann universal constructor:

The mathematician John von Neumann proposed a cellular automata model of a universal constructor capable of self-replication. Many researchers have refined and improved von Neumann's original design since it was first presented in the 1950s. The original design used tens of thousands of cells in a two-dimensional CA with 29 states. The general structure of the model is a movable constructing arm controlled by instructions encoded in a long line of cells resembling a tape, conceptually similar to a computer-controlled robot arm.

This work was made possible by support from the Royal Thai government, the Mellon Foundation, and the National Science Foundation under GrantITS 0205466. The results and opinions expressed are solely those of the authors. We thank A. Cushing, Y. Kwon, K. Lee, M. Kutzer and all of theundergraduate students in the Mechatronics courses over the past five years who have designed several of the prototypes shown in this review. The workpresented in Sect. “Towards a Universal Constructor” was performed at University of NewMexico with help from Professors G. Starr, H. Tran, and J. Wood.

Authors and Affiliations

1. Biomedical and Robotics Technology Lab, Faculty of Engineering, Mahidol University, Salaya, Thailand

   Jackrit Suthakorn

2. Robot and Protein Kinematics Lab, Department of Mechanical Engineering, Johns Hopkins University, Baltimore, USA

   Matthew Moses & Gregory S. Chirikjian

Editors and Affiliations

1. RAMTECH LIMITED, 122 Escalle Lane, Larkspur, CA, 94939, USA

   Robert A. Meyers Ph. D. (Editor-in-Chief) (Editor-in-Chief)

Reprints and permissions

© 2009 Springer-Verlag

Policies and ethics