| Article ID | Journal | Published Year | Pages | File Type |
|---|---|---|---|---|
| 412154 | Robotics and Autonomous Systems | 2010 | 8 Pages |
One of the captivating characteristics of social insects, in spite of their rudimentary individual constitution, is the ability by which they have to solve complicated problems in an elastic and robust way. This includes elasticity which ensures the adaptation of the insect system to the unpredictable changes of their environment, and robustness which guarantees a functioning continuity of the system, in spite of the possible failure of a certain number of its elements in the achievement of their individual missions. From this point of view, fields of research have emerged over the past decades, with the aim of trying to reveal the secret behind the relationship between individual and society, so perfectly designed in nature. Collective robotics is one of those fields where we try to find microscopic rules allowing a group of autonomous robots, mobile and with limited capacity, to carry out a specific macro-task, such as exclusive positioned heap formation. The idea, behind this, is to use a model of oriented reactive agent simulation, to seek the relations which can link the local perceptions of the simulated robots with their basic actions, in order to make the above mentioned gathering task a success. An evolutionary approach is used for this purpose, making it possible to discover the functional control relations of these simulated robots. An analogy with the precepts specific to the ant community is established and results of simulation indicating the effectiveness of the detected rules are presented.
Research highlights► Strategies guiding robots to succeed an exclusive positioned grouping are found and validated. ► They belong to a class where the structure is defined by using properties of real robots. ► They are deduced from rules discovered in a simulation environment by using a reactive agent model. ► These rules are an optimal solution of an evolutionary research process based on the reverse emergence. ► The found strategies are validated by using a scaling and an analogy with those detected in ants.
