Swarm Intelligence: From Natural to Artificial Systems
Published Jun 1, 2002 · B. Webb
Connection Science
3,144
Citations
222
Influential Citations
Abstract
‘Swarm intelligence’ is defined in the preface of this book to be ‘the emergent collective intelligence of groups of simple agents’. As such, there is a huge range of work that the book could cover – from distributed software agents to cellular automata. A more accurate gloss of the actual content of the book comes in the introductory chapter – ‘designing problem solving devices inspired by social insects’ (p.23). Within this more circumscribed eld, the book provides fairly comprehensive coverage of recent research developments. It thus constitutes an excellent resource for researchers in the eld or for those wishing to familiarize themselves with current approaches e.g. it would be an ideal introduction for a doctoral student wanting to enter this area. But why and how should social insects inspire methods for problem solving? The introduction (Chapter 1) claims the critical concepts are self-organization and stigmergy. Self-organization is more or less self-explanatory – the idea that some kind of global coherence in structure or behaviour can emerge from purely local interactions. As such it is not a new concept (as the authors recognize, much of the theoretical work in this area comes from physics) nor is it evident that it applies to all social insect behaviour. For example, it could be argued that the idea (discussed in Chapter 5) of a template – a structure in the environment that organises the local activity towards a constrained result – is a global mechanism in much the same way as a centralized plan. However it is true that much of the success and exibility of social insect behaviour does seem to be the emergent result of local actions. Insect swarms provide an existence proof of the potential power of self-organizing systems. They can also provide specic ideas as to the kinds of local interactions to implement. If termites can build a huge, well-structured nest, and ants alternate their activities dynamically to increase colony efciency, then it should be possible to copy their local behaviours to achieve comparable global results. Stigmergy may be a less familiar term but refers to one important mechanism for successful self-organization: how agents’ effects on their environment can feed back to modify each others’ and their own behaviour. A good example discussed at length in Chapter 4 is distributed clustering. A simple agent behaviour of picking up isolated objects and dropping them when another object is encountered leads to the formation of initial small clusters that can effectively guide the further development of larger ones. These ideas have become an attractive (some might say seductive) alternative to more conventional ‘centralized’ approaches to problem solving. By providing an overview, this book allows some assessment of how successful this approach has been, as discussed further below. The presentation is nicely structured, with each chapter taking a different example of insect social behaviour – foraging, task allocation, brood sorting, self-organization and templates, nest building and co-operative transport – and presenting rst the biological data, then models proposed to explain it, then the generalization of such models to solve analogous problems such as routing in Connection Science, Vol. 14, No. 2, 2002, 163–164
Sign up to use Study Snapshot
Consensus is limited without an account. Sign up or log in to use Study Snapshot and unlock more functionality.