OMENA, R. A. L. V.; http://lattes.cnpq.br/6549291140653300; OMENA, Rômulo Afonso Luna Vianna de.
Resumen:
Automated Guided Vehicles (AGVs) are essential for industry material transportation. In
the Industry 4.0 and Industrial Internet of Things scenario, the AGV fleet is expected to be
connected and integrated into the factory management system, being flexible and adapting
to new demands. AGV control systems with fixed path navigation may not meet these re-
quirements. Edge computing brings cloud resources to the network’s edge, making them
closer to users. These resources can be accessed through a wireless network and applied to
industrial demands. The AGVs can benefit from this when offloading tasks that require more
computing resources to the edge server. However, the wireless network in the industrial en-
vironment is subject to degradation due to interference, signal reflections, shadowing effects,
and electromagnetic wave absorption, among other challenges. The AGV, as a mobile robot,
may traverse areas where the signal is degraded, increasing risks of collisions and accidents.
Results of experiments suggest that Model Predictive Control (MPC) executed at the edge
server, combined with a delay and packet loss compensation strategy implemented in the
robot, can be used to mitigate these network degradations. In sequence, a two-tier architec-
ture with MPCs is proposed to control multiple AGVs. The first tier, executed on the edge
server, plans the trajectory of the AGVs globally, preventing collisions of the AGVs with
fixed obstacles and each other. In the computer embedded in the AGV, the compensator used
in the previous experiments gives place to a trajectory-tracking MPC, which must receive the
trajectory of the respective AGV from the edge server and track it. Results of experiments
carried out in four validation scenarios indicate that from the proposed architecture, it is pos-
sible to drive the AGVs without collisions, even in the communication network’s occurrence
of delays and packet losses. In addition, tasks that demand more computational resources
are offloaded to the edge server so that the computer embedded in the AGV can have more
restricted resources, reducing costs and battery consumption