AGV navigation in complex environments: A cooperative trajectory optimization strategy based on SMF-MPC
TL;DR
A set-membership filter algorithm that constructs a minimum enclosing ellipsoid to enclose the system’s true state in real time and incorporates the geometric characteristics of this ellipsoid as a time-varying penalty term into the MPC optimization objective, thereby achieving dynamic risk avoidance.
Abstract
This paper addresses the navigation problem of automated guided vehicles (AGVs) in complex environments by proposing a collaborative trajectory optimization strategy based on the Set Membership Filter (SMF) and Model Predictive Control (MPC). In contrast, the traditional probability distribution-based navigation method, UKF-MPC, often fails to provide a strict safety margin when dealing with UBB noise, thereby increasing the risk of navigation failure. Therefore, this paper proposes a set-membership filter algorithm that constructs a minimum enclosing ellipsoid to enclose the system’s true state in real time and incorporates the geometric characteristics of this ellipsoid as a time-varying penalty term into the MPC optimization objective, thereby achieving dynamic risk avoidance. Finally, comparative experiments under three different types of environmental noise were conducted using MATLAB. The MATLAB simulation results indicate that, under complex environmental disturbances, the SMF-MPC control strategy exhibits a higher navigation success rate and greater robustness than both the traditional UKF-MPC control strategy and the Robust MPC control strategy.