Multi-Layered Multi-Robot Control Architecture for the Robocup Logistics League

摘要

For many years, different architectures for controlling robotic systems have aroused with the goal of connecting the real world with the deliberation capabilities of the artificial system. In most cases, different modules are defined and connected to make data to flow from sensors to deliberation, and commands from deliberation to actuators. In such path, the different modules implement hierarchical levels, abstractions or layers that are traversed. However, in most cases, such modules are defined specifically for the architectures, and reusing them is typically very difficult or even unfeasible. In this work, we propose a layered architecture based on a specific set of modules, which could be replicated and reused in many different circumstances. The architecture is based on the use of automated planning, although any other deliberation mechanism could be used. We demonstrate its usability in the Robocup Logistics competition, and report a first implementation ready to compete in a future edition.