From Bio-Inspired to Institutional-Inspired Collective Robotics

From ISRWiki
Jump to navigation Jump to search

Abstract

We seek to study and formalise laws that govern collective systems with the aim of synthesising systems of relatively simple robots that display complex behaviour individuals through local interactions. In order to achieve this endeavour, we will study both biological systems and social systems. From biology, we will focus on cell populations. A single cell is relatively simple when compared with a cell population such as a biological creature. While from sociology, we will focus on institutional economics (at the macro level, the agents can be treated as simple entities, no matter how sophisticated they may be at the micro level). Our objective is to bring together theories, ideas and inspiration from institutional economics and cell biology under a common formal framework for large robot populations modelling and analysis.

In recent years, the predominant inspiration for collective robotics modelling analysis and design has been originated from biology. In the case of swarm robotics (SR), inspiration is taken from studies of the self-organizing capabilities displayed by social insects such as ants. Drawing inspiration from these types of natural systems, SR concepts have been applied to swarms of robots with limited sensing and actuation capabilities, performing relatively simple cooperative tasks such as foraging, coverage or odour tracking.

Despite the reasonable success of SR in relatively simple applications, there is no systematic method to design individual behaviours at the micro level, including their interaction actions for the desired collective behaviour to at the macro level to emerge; in fact, the emergent nature of the collective behaviour is a principle that precludes goal- or performance-oriented design. Economics-inspired experiences within multiagent systems suggest that merely emergent processes, and simple local interactions between individuals sometimes lead to inefficient solutions to collective problems (Caldas 2001).

Economics can provide some insights on how to deal with large collective systems. Institutional Economics (IE) [e.g., Hodgson 2000] takes institutions – coordination devices deliberately set up by agents or evolved out of interaction – as key elements of any sophisticated society. Notwithstanding the lessons we can learn from IE, its opposition to “naturalistic” approaches can lie in the way to a unified (bio- and institutional-inspired) framework to the analysis and synthesis of collective systems.

Considering other domains in the biological landscape can be of help here. In biology, epigenetics refers to heritable phenotypic variation without genetic variation. An epigenetic approach can provide a new common ground for genetic and non-genetic views on the micro-macro link problem. Our working hypothesis regarding epigenetics is that collective systems have a state space with multiple stationary states, where the actual repertoire of inputs and outputs displayed by each agent (cell or robotic system) in a given instant, as well as the way in which these are wired within the agent (gene regulatory network in cells, and software/hardware in robotic systems), can be different even in similar or identical environments. On the basis of this hypothesis, new methods for controlling collectives, both at micro and macro levels, can be developed.

By the end of this project, we will aim at providing novel cooperation and coordination algorithms and methods for collective robotics based on a formal framework for collective systems. More specifically, we will import concepts from Biology and Economy and develop probabilistic mathematical models of the robot population dynamics under the different approaches (bio- and institutionally-inspired, and merged). This will allow us to gain further insight on their differences and similarities, relative merits, and adequateness for particular classes of systems, by predicting their performance in non-tested scenarios. Furthermore, though the project focus is on bringing multidisciplinary inspiration to design better robot collectives, we expect to provide further insight on cell biology and institutional economics mathematical modeling as a side effect. The three core investigators in the project have strong experience in, respectively, modelling and control of multirobot systems, the genetic basis of development and evolution of complex systems, and swarm robotics systems. Previous work from the principal researchers has paved the way for the proposed research. Pedro Lima and Jorge Carneiro co-supervised the PhD of Dejan Milutinovic on stochastic hybrid automata models of cell population dynamics, as well as modelling and control of robot swarms [MCAL07, DL07, DL06]. Porfírio Silva and Pedro Lima have introduced the institutional robotics concept in Porfírio Silva’s PhD thesis and two conference papers, one of them [SL07] awarded the Best Philosophy Paper Award at ECAL 2007.

Resumo / Robótica Colectiva: Da Inspiração Biológica à Inspiração Institucionalista


News and Events

Partners

People

  • INESC Porto:
    • António Paulo Gomes Mendes Moreira
    • Luis Miguel Pinho de Almeida
    • Fernando Arménio da Costa Castro e Fo...
    • Paulo José Cerqueira Gomes da Costa
    • Amélia Cristina Duque Caldeira Matos
    • Dalila Benedita Machado Martins Fonte...
  • ISEP/IPP
    • Eduardo Alexandre Pereira da Silva
    • Alfredo Manuel de Oliveira Martins
    • André Miguel Pinheiro Dias
    • Hugo Miguel Gomes da Silva
    • José Miguel Soares de Almeida

Reports

(not yet available)

Media

Publications

  • Cooperative Localization Based on Visually Shared Objects, P. Lima, J. Santos, P. Santos, R. Oliveira, A. Ahmad, Proc. of RoboCup2010 Symposium, Singapore, 2010

Links

Reference Information

  • Project Reference: PTDC/EEA-CRO/104658/2008
  • Start: January 2010
  • End: December 2012
  • Total budget: 197.976,00 EUR
  • Principal Investigator: Pedro U. Lima
Retrieved from "http://mediawiki.isr.tecnico.ulisboa.pt/index.php?title=From_Bio-Inspired_to_Institutional-Inspired_Collective_Robotics&oldid=2760"