Don't stamp on social insects

This article is taken from a 2009 blog which explored topics in science and technology.  It remains pertinent today due to the high media profile of robotics and autonomous machines.

Reading time: 3 mins



‘Social insect’ species, such as ants, bees and termites have recently become the focus of attention from systems analysts and operational researchers. Their interest stems from the fact that insect societies are subject to no centralised or hierarchical decision making, and yet they're synonymous with the construction of sophisticated nests (such as termite mounds), efficient food gathering, and other impressive organisational feats.

Termite mounds are built without project managers
or Gantt charts (or should that be ant charts?)

French physicist, Eric Bonabeau, explained in the Harvard Business Review (2001) that social insect species are successful because of their flexibility (adapting to changing environments), robustness (ability to perform collectively when one or more individuals fail), and self-organising characteristics (neither centrally controlled nor locally supervised).

This so-called ‘swarm intelligence’ is based on simple, low-level communications between the insects. The challenge for technologists is to exploit these properties, developing practical solutions to human organisational problems (research goals which could be viewed as a form of reverse anthropomorphism).

A number of real-world swarm applications have already been deployed, while other technically-viable solutions are at advanced stages of planning or development, including:

• truck despatch algorithms for Air Liquide in France;
• optimisation of cargo distribution for Southwest Airlines;
• research into pilotless aircraft swarms for the US Air Force;
• use of clustering algorithms for commercial databases in order to identify trends and commonalities;
• development of highly effective data networks using efficient paths selected in real time.

Is the work of Bonabeau and others really introducing a new technological and organisational paradigm, or has the social insect metaphor become a convenient hook on which scientists have merely chosen to hang a number of current interests? Dusan Teodovoric, a researcher engaged in transport modeling, suggests that we may be seeing more of the latter:

“It is important to state here that the development of artificial systems does not entail the complete imitation of natural systems, but explores them in search of ideas and models.” (Teodovoric, 2003)

Similarly, in the conclusion to their paper on ‘ant colony optimization’ (or ACO), Dorigo and Di Caro (1999) state that:

“As is very common in the practical usage of these heuristics, ACO algorithms often end up at some distance from their inspiring natural metaphor. Often ACO algorithms are enriched with capacities that do not find a counterpart in real ants…”

Their enrichment of the social insect metaphor with non-insect characteristics, such as memory of previous actions and complex inter-agent communications, suggests that the entomological model may be incapable of independently supporting useful applications.

We might also question the metaphor’s novelty. Do ants really provide new and unique insights? It has long been argued that human beings are guided, like insects, by a small number of simple rules: economists, for example, talk about rationality and utility-maximising behaviour, while Freud focused on reproductive urges. These perspectives may, in turn, be used to explain complex emergent phenomena, such as financial markets and gender relations.

Human population centres emerge from individual decisions about where to obtain secure supplies of food and water (i.e. where to settle). Therefore, every city, as with every termite mound, could be said to emerge from very simple, instinctive, low level multi-agent decision-making.

Nevertheless, regardless of the novelty or true origins of these ideas, there remains a possibility that the social insect paradigm will serve humanity through new and better technologies. If so, we should support the work of Bonabeau and others.

Self-organising miniature robots

The possibilities are intriguing, if slightly unsettling. The photograph above shows a robot swarm organising themselves in a lab at the University of Stuttgart. Without doubt, an intelligent swarm of cleaning machines at home would greatly improve my quality of life. On the other hand, self-organised nano-scale robots silently and invisibly serving my perceived ‘needs’ may be a less attractive prospect. What exactly will they be doing and how would I know?

Perhaps there is a moral here. Next time you see an ant, overcome the urge to stamp on it. Simplistic or instinctive responses are not always a force for good. It would be sensible to treat real social insects as you would like the autonomous machines of the future to treat you!



Writer: PJ Moar of Moar Partnerships
Email: p.moar@moar.com
Twitter: @MoarPart