Search for a command to run...
This paper has presented the simplest solution so far to the problem of clustering objects with a swarm of robots. The robots are memory-less and are unable to perform arithmetic computations. They are only able to detect the presence of an object or another robot in their line of sight. As they do not possess any distance information about a perceived object (nor any force sensors), they are unable to tell whether they are manipulating it. Despite these limitations, we identified a controller that is capable of consistently gathering the objects into a single cluster. Simulation results have shown that the ability of the robots to distinguish between objects and other robots is beneficial; indeed, if the robots can only perceive other robots as objects, the behavior does not scale well with increasing numbers of robots. Simulations have also shown that the controller is fairly robust with respect to sensory noise. The sensor/controller solution was implemented on a physical system of 5 e-puck robots, and favourable results have been obtained in systematic experiments with 20 objects, with 86.5% of the objects being gathered into one cluster after 10 minutes (average across 10 trials). In the future, we intend to implement a solution of the object clustering task with robots at the sub-millimeter scale.