Report
Reference
Social Amoeba Dictyostelium Discoideum as an Inspiration for Engineering of Swarm Robotics
PARHIZKAR, Mohammad
Abstract
Understanding the collective behaviors in nature and its potential links to engineering the collective artificial behaviors in swarm robotics have attracted the attention among researchers. They have various impacts on different domains such as cell-biology, cancer study, a swarm of drones and unmanned robots. Since the cancer cells share similar collective behaviors, the biomedicine researchers look into different examples from nature to design anti-cancer drugs to shrink tumors in human bodies. An interesting form of collective system is demonstrated by Dictyostelium discoideum. The Ph.D. thesis project aims in the study and understanding of different phases of D. discoideum1 from simple computational models to the concept of multi-agent systems and swarms robotics.
PARHIZKAR, Mohammad. Social Amoeba Dictyostelium Discoideum as an Inspiration for Engineering of Swarm Robotics . 2016
Available at:
http://archive-ouverte.unige.ch/unige:116285
Disclaimer: layout of this document may differ from the published version.
1 / 1
Social Amoeba Dictyostelium Discoideum as an Inspiration for Engineering of Swarm Robotics
Mohammad Parhizkar Information Systems Geneva School of Social Sciences Centre Universitaire d’Informatique Superviser: Prof. Giovanna Di Marzo Serugendo
Abstract
Understanding the collective behaviours in nature and its potential links to engineering the collective artificial behaviours in swarm robotics have attracted the attention among researchers. They have various impacts on different domains such as cell-biology, cancer study, swarm of drones and unmanned robots. Since the cancer cells share similar collective behaviours, the biomedicine researchers look into different examples from nature to design anti-cancer drugs to shrink tumours in human bodies. An interesting form of collective system is demonstrated by Dictyostelium discoideum.
The Ph.D. thesis project aims in the study and understanding of different phases ofD. discoideum1 from simple computational models to the concept of multi-agent systems and swarm robotics.
Keywords: Dictyostelium discoideum - Swarm robotics - Multi-agent systems - Self-organising systems
Thesis Problem
In the last decade, bio-inspired multi-agent systems have become increasingly useful for the design of swarm of robots. In the term of swarm intelligence, there are lot of surveys on social insects, such as: “nest building in termite”, “honeybee societies”, “foraging in ant colonies”, “fish schooling” and “bird flocking”. Besides them, another well-known example is D. discoideum (slime mold) with the ability of self-aggregation, dynamic self- assembly and self-disassembly, which play a significant role in their development process, to build a complex multicellular organisms.
D. discoideum is an eukaryote that is related to animals and fungi. It is a social amoeba, that feeds on bacteria in the top few centimetres of soil and multiplies by binary fission. The striking feature of Dictyostelium cells is that they undergo a relatively simple programme of multicellular development, which in many ways resembles animal development. This multicellular development process helps them to switch behaviour to survive in the lack of food: individual cells move around on their own when there is plenty of food (vegetative phase); when food is scarce, cells use chemical gradient (namely cAMP) to build a coherent and cohesive organisms , similar to a slug (approximately with the same size contains 2×104 to 1×105 cells) moving in a coordinated way towards areas with more heath and more light. Then they transform to a new organism called fruiting body consisting of a globule of spore cells and a slender stalk. Its function is to hold the spore mass as high off the ground as possible, for optimal spore dispersal.
D. discoideum life cycle as an excellent example of emergent, inspires us to investigate the relation between first-order and higher-order collective behaviours in terms of emergence. Second-order emergent behaviour arises from the interactions of individuals, which are themselves the result of first-order emergent societies. Second- order emergence, refers to systems in which agents recognise the existence of groups that emerged from their own collective behaviours.
The main objectives of this project is to: (1) provide agent-based models of the different phases of D.
discoideum life cycle, (2) extract pertinent mechanisms for higher-order emergent behaviour and provide them as design patterns for artificial systems; and (3) eventually translate these mechanisms into swarms of real multi-modular, self-configuring, self-adaptive micro-robots, that we call DictyRobots. This project therefore involves the combination of different disciplines - cell biology; self-organising systems; swarm intelligence; and swarm robotics - into one activity. DictyRobotproject addresses these points and tackles the following research questions:
• What are the social relations and configurations ofD. discoideum behaviours at the different phases of its life cycle and how to model them?
• What are the mechanisms favouring higher-order emergence in swarms and artificial collective behaviour?
• How to translate and implement those mechanisms into a swarm of robots (real robots / simulated agents)?
DictyRobotproject will substantially advance the state of the art by providing:
• Fine-grained understanding ofD. discoideum individual cells behaviours at all phases of its life cycle and provision of corresponding agent-based models validated with actual biological experiments;
• Novel self-organising mechanisms for higher-order emergent behaviours, expressed and defined as design patterns for artificial systems;
1Dictyostelium discoideum
1
• Translation of first-order and higher-order emergent behaviours into swarms of real multi-modular, self- configuring, self-adaptive micro-robots, that we call “DictyRobots”.
Ph.D. Plan
The whole Ph.D. project consists of four major steps, each spanning approximately one year. The main objective during each phase is to propose and evaluate several models and have a comparison at the end as a conclusion.
Phase 1:Understanding and modellingD. discoideum first-order emergent behaviour Phase 2:Translating D. discoideum first-order collective behaviour to swarms of robots Phase 3:Understanding and modellingD. discoideum higher-order emergent behaviour Phase 4:Translating D. discoideum higher-order collective behaviour to multi-structure robots
Plan for Next Step
During the first two years ofDictyRobotsproject we worked on “Phase 1”. We studied different phases of the life cycle and implemented various models for aggregation and streaming phases. During the next year we would like to work on the second phase, which includes two major tasks:
• Simulations for the rest ofD. discoideumlife cycle
• Setting up a visit to the University of Graz to translate represented model on the real swarm of robots.
This step will be done by collaboration of Prof. Thomas Schmickl and his group2.
Then, we submitted another proposal for ‘SNSF funding project’ and we are still waiting for their answer.
Achieved Results
Developing different agent-based models and simulations for the early phases of D. discoideum life cycle (first- order emergence). These simulations include the following items:
• An agent-based model for vegetative phase to mimic the production and consumptions of bacteria and also individual amoebae. Using ‘Java Repstset-simphoney’ programming language, we designed and developed an individual based model to simulate the growth and behaviour of both amoebae and bacterial populations.
Since our work is a decentralised and individual based model, we have constructed our simulation code in terms of individual amoebae cells and their behaviour, within a two-dimensional spatial grid as the environment. In this model, individual amoeba senses the highest concentration of the nutrient (bacteria) around itself and moves toward it. The relationship between the rate of growth and consumption for both two bacterial population and amoebae population depends of the density and initial number of agents in the field.
• A model for aggregation phase includes stream formation and aggregation territories using positive feed- backs. This model is discrete in space and time, and controls a group of amoebae cells at each time step, using a set of random variables for each agent . This model is for individual and collective cell movement toward the aggregation centers including the cAMP concentration, cAMP diffusion, cell-cell signalling, signal transduction, PDE affects, signal pattern, stream formation and cell motility.
Publications
After validating our simulation results with “Prof. Thierry Soldati” from “Biochemistry Department” for com- paring of the simulation output with observed aggregation patterns, we plan to publish a paper on our study in
“IC-SI”, during March 2017. In that paper, we would like to investigate different models of aggregation phase as the first step of D. discoideum life cycle. These simple models will be developed describing how the behaviour of individual d. discoideum cells interacts to cause the patterns seen in aggregation. Our results of the simulations describe how some factors affect wave propagation and centre behaviour during aggregation of D. discoideum The following two publications are available:
1. A progress report providing a unified survey of the relevant literature aiming at understanding D. dis- coideum behavour. It covers the following aspects: a. Dictyostelium different species with different pat- terns; b.D. discoideumgenomics and molecular genetics; c. Different phases of morphogenesis and different emergent properties: Vegetative, aggregation, migration, culmination, growth; d. Different periodic chem- ical cell-cell signals propagations: Positive feedback loop; e. Chemotaxis: The directed movement in a gradient of chemoattractant; f. Motility and adhesion: Extension of pseudopodia; g. Quorum sensing: A system to coordinate tasks among cells according to the density; h. Spiral and concentric waves before the multicellularity; i. Cell determination and differentiation in different phases of the development.
2. Presenting D. discoideum as a bio-inspired model for swarm robotics is a creative idea. Description of research questions and preliminary simulation results on aggregation territories are published in ourSASO - 2015 paper.
2Artificial Life Laboratory - Department for Zoology - University of Graz, Austria
2
