Autonomous self-repair systems : A thesis submitted in partial fulfilment of the requirements for the Degree of Doctor of Philosophy at Lincoln University
Citations
Altmetric:
Authors
Date
2021
Type
Thesis
Collections
Keywords
self-repair, regeneration, Homeostasis, bioelectrical homeostasis, anatomical homeostasis, multi-cellular structures, planaria, collective intelligence, computational tissues, somatic cell network, perceptrons, neural networks, signalling entropy, information fields, auto-associative neural networks, robotics
Abstract
Regeneration is an important and wonderful phenomenon in nature and plays a key role in living organisms that are capable of recovery from trivial to serious injury to reclaim a fully functional state and pattern/anatomical homeostasis (equilibrium). Studying regeneration can help develop hypotheses for understanding regenerative mechanisms along with advancing synthetic biology for regenerative medicine and development of cancer and anti-ageing drugs. Further, it can contribute to nature-inspired computing for self-repair in other fields. However, despite decades of study, what possible mechanisms and algorithms are used in the regeneration process remain an open question. Therefore, the main goal of this thesis is to propose a comprehensive hypothetical conceptual framework with possible mechanisms and algorithms of biological regeneration that mimics the observed features of regeneration in living organisms and achieves body-wide immortality, similar to the planarian flatworm, about 20mm long and 3mm wide, living in both saltwater and freshwater. This is a problem of collective decision making by the cells in an organism to achieve the high-level goal of returning to normality of both anatomical and functional homeostasis. To fulfil this goal, the proposed framework contains three sub-frameworks corresponding to three main objectives of the thesis: self-regeneration or self-repair (anatomical homeostasis) of a simple in silico tissue and a whole organism consisting of these tissues based on simplified formats of cellular communication, and an extension to more realistic bioelectric communication for restoring both anatomical and bioelectric homeostasis.
The first objective is to develop a simple tissue model that regenerates autonomously after damage. Accordingly, we present a computational framework for an autonomous self-repair system that allows for sensing, detecting and regenerating an artificial (in silico) circular tissue containing thousands of cells. This system consists of two sub-models: Global Sensing and Local Sensing that collaborate to sense and repair diverse damages. It is largely a neural system with a perceptron (binary) network performing tissue computations. The results showed that the system is robust and efficient in damage detection and accurate regeneration.
The second objective is to extend the simple circular tissue model to other geometric shapes and assemble them into a small virtual organism that regenerates similar to the body-wide immortality of the planarian flatworm. Accordingly, we proposed a computational framework extending the tissue repair framework developed in Objective 1 to model whole organism regeneration that implemented algorithms and mechanisms to achieve accurate and complete regeneration in an (in silico) worm-like organism. The system consists of two levels: tissue and organism levels that integrate to recognise and recover from any damage, even extreme damage cases. The tissue level consists of three tissue repair models for head, body and tail. The organism level connects the tissues together to form the worm. The two levels form an integrated neural feedback control system with perceptron (binary) for tissue computing and linear neural networks for organism-level computing. Our simulation results showed that the framework is very robust in returning the system to the normal state after any small or large scale damage.
The last objective is to extend the whole organism regeneration framework developed in Objective 2 by incorporating bioelectricity as the format of communication between cells to make the model better resemble living organisms and to restore not only anatomy but also basic functionality such as restoring body-wide bioelectric pattern needed for physiological functioning in living systems. We greatly extended the second framework by conceptualising and modelling mechanisms and algorithms that mimicked both the pattern and function restoration observed in living organisms and implemented it on the same artificial (in silico) organism developed in Objective 2 but with greater realism of the anatomical structure. This proposed framework consists of three levels that collaborate to fully regenerate the anatomical pattern and maintain bioelectric homeostasis in the in silico worm-like organism. These three levels represent tissue and organism models for regeneration and body-wide bioelectric model for restoring bioelectric homeostasis, respectively. They extend the previous neural feedback control system to integrate another (3rd) level, bioelectric homeostasis. Our simulations showed that the system maintains and restores bioelectric homeostasis accurately under random perturbations of bioelectric status under no damage conditions. It is also very robust and plastic in restoring the system to the normal anatomical pattern and bioelectric homeostasis after any type of damage.
Our framework robustly achieves some observations of extreme regeneration of planaria like body-wide immortality. It could also be helpful in engineering for building self-repair robots, biobots and artificial self-repair systems.
Permalink
Source DOI
Rights
https://researcharchive.lincoln.ac.nz/pages/rights