Extracted from New Scientist 16 August 2008
“Given a choice between two different flower beds, how can honeybees hunting for nectar be sure they’ve chosen the best patch? A new computer model may provide the answer, as well as insights into the workings of a “hive mind” that could be used to guide swarms of robots.
To test this hypothesis, Ronald Thenius of the University of Graz in Austria built a computer simulation of a hive containing 5000 independent virtual bees. Each forager started out visiting one of two different flower patches, but would switch destinations if it had to wait too long to be unloaded or was being serviced by too many receivers.
The idea that bees glean information from the number of unloadings is new, says Francis Rietnieks, a bee expert at the University of Sussex in the UK, but it needs to be verified in the field. “If their simulation suggests a novel means of information transfer, ideally they will devise a suitable experiment that can test the model’s predictions,” he says.
Thenius says the work could prove useful in controlling swarms of tiny robots for sensing and surveillance applications. Such robots could use a similar method of incidental communication to arrive at group decisions that could maximise resources. The system would be robust because it would rely on very simple observations, he notes.
The results, presented at the Artificial Life IX conference in Winchester, UK, last week were promising. The virtual bees moved to the better nectar source at similar rates and in similar proportions to those observed for real bees. “It’s like a new pub has opened with cheap beer: everyone’s trying to find it,” says Thenius. “The hive can gain up to 20 per cent more nectar this way.”