Hierarchical self replication

Abstract

This dissertation focuses on Robotic Self-Replication. The development of an autonomous self-replicating mobile robot that functions by undergoing stochastic motions is presented. An associated statistical analysis is presented to quantify the behavior of this system. The robot functions hierarchically. There are three stages in this hierarchy: (1) an initial pool of feed modules/parts together with one functional basic robot; (2) a collection of basic robots that spontaneously forms out of these parts as a result of a chain reaction induced by stochastic motion of the initial seed robot in stage 1; (3) complex formations of joined basic robots from stage 2.

In the first part we demonstrate basic stochastic self-replication in unstructured environments. A single functional robot moves around at random in a sea of stock modules and catalyzes the conversion of these modules into replicas. In the second part of the thesis, the robots are upgraded with a layer that enables mechanical connections between robots. The replicas can then connect to each other and aggregate. Robotic assembly for the replicas can take place through random, semi-random and programmable formations. The number of possible outcomes is studied.

Pseudo-dynamical and stochastic simulations are presented and compared with results from physical experiments. After verifying that the simulations produce the same results as the real experiments, we apply them to large populations to extrapolate how the behavior of real systems might scale.