RT info:eu-repo/semantics/article T1 Combining reinforcement learning and conventional control to improve automatic guided vehicles tracking of complex trajectories A1 Sierra Garcia, Jesús Enrique A1 Santos, Matilde K1 Automated guided vehicle (AGV) K1 Intelligent control K1 Machine learning (ML) K1 Path following K1 PDI K1 Reinforcement learning (RL) K1 Ingeniería eléctrica K1 Electric engineering AB With the rapid growth of logistics transportation in the framework of Industry 4.0,automated guided vehicle (AGV) technologies have developed speedily. These systems present two coupled control problems: the control of the longitudinal velocity,essential to ensure the application requirements such as throughput and tag time,and the trajectory tracking control, necessary to ensure the proper accuracy in loading and unloading manoeuvres. When the paths are very short or have abruptchanges, the kinematic constraints play a restrictive role, and the tracking controlbecomes more challenging. In this case, advanced control strategies such as thosebased on intelligent techniques, including machine learning (ML) can be useful.Hence, in this work, we present an intelligent hybrid control scheme that combinesreinforcement learning-based control (RLC) with conventional PI regulators to faceboth control problems simultaneously. On the one hand, PIs are used to control thespeed of each wheel. On the other hand, the input reference of these regulators iscalculated by the RLC in order to reduce the guiding error of the path tracking and tomaintain the longitudinal speed. The latter is compared with a PID path followingcontroller. The PID regulators have been tuned by genetic algorithms. The RLC allowsthe vehicle to learn how to improve the trajectory tracking in an adaptive way andthus, the AGV can face disturbances or unknown physical system parameters thatmay change due to friction and degradation of AGV mechanical components. Extensive simulation experiments of the proposed intelligent control strategy on a hybridtricycle and differential AGV model, that considers the kinematics and the dynamicsof the vehicle, prove the efficiency of the approach when following differentdemanding trajectories. The performance of the RL tracking controller in comparisonwith the optimized PID gives errors around 70% smaller, and the average maximumerror is also 48% lower. PB Wiley SN 0266-4720 YR 2022 FD 2022-06 LK http://hdl.handle.net/10259/7408 UL http://hdl.handle.net/10259/7408 LA eng NO Open access funding enabled and organized by Projekt DEAL. DS Repositorio Institucional de la Universidad de Burgos RD 28-mar-2024