One of the greatest challenges in robotics is to create machines that are a
ble to interact with unpredictable environments in real time. A possible so
lution may be to use swarms of robots behaving in a self-organized manner,
similar to workers in an ant colony(1-5). Efficient mechanisms of division
of labour, in particular series-parallel operation and transfer of informat
ion among group members(6), are key components of the tremendous ecological
success of ants(7,8). Here we show that the general principles regulating
division of labour in ant colonies indeed allow the design of flexible, rob
ust and effective robotic systems. Groups of robots using ant-inspired algo
rithms of decentralized control techniques foraged more efficiently and mai
ntained higher levels of group energy than single robots. But the benefits
of group living decreased in larger groups, most probably because of interf
erence during foraging. Intriguingly, a similar relationship between group
size and efficiency has been documented in social insects(9-11). Moreover,
when food items were clustered, groups where robots could recruit other rob
ots in an ant-like manner were more efficient than groups without informati
on transfer, suggesting that group dynamics of swarms of robots may follow
rules similar to those governing social insects.