Task partitioning

by Giovanni Pini, Arne Brutschy, Alexander Scheidler, Mauro Birattari, and Marco Dorigo
DATE TODO

This page reports data collected in the experimental runs performed with the real robots. We performed a total of 12 experimental runs, in 6 of these runs the robots employ the non-partitioning strategy, in the remaining 6 the robots employ the partitioning strategy.

For each run we report basic statistics:

• Number of objects retrieved: amound of objects delivered to the nest by the swarm, at the end of the experiment
• Get lost frequency, objects source: frequency at which local search failed or the robots attempted to exit the arena. The measure accounts for the cases in which the robots were trying to reach the source (i.e., the source position estimate is relative to the objects source).
• Get lost frequency, transfer locations: frequency at which local search failed or the robots attempted to exit the arena. The measure account for the cases in which the robots were trying to reach the position in which they received an object (i.e., the source position estimate is relative to the location where the robot received an object).
• Number of time robots tried to exit the arena: how many times the mechanism that checks if the robot tries to exit the arena was triggered.
• Number of object transferred: number of time an object was passed from a robot to another (partitioning strategy only).
• Number of time the robots abandon transferring (timeout): how many times robots waiting for transfer their object timed out (partitioning strategy only).
• Abandon transfer rate: rate at which the robots waiting for tranferring the object abandoned (partitioning strategy only).
• Number of objects lost: number of times an object was dropped on the floor before reaching the nest (object dropped are removed from the experiment).