Work

Set-membership estimator-based proximal policy optimization algorithm for trajectory tracking control of automated guided vehicles

Year: 2025

Type: article

Abstract: When an automated guided vehicle follows a desired trajectory, the precision of the trajectory tracking results may be inevitably affected by unknown-but-bounded noises, such as the ground friction, battery fluctuations, and other factors. Thus, this paper investigates the trajectory tracking problem of automated guided vehicles subject to such noises. First, a proximal policy optimization algorithm based on a set-membership estimator is proposed to realize the trajectory tracking of automated guided vehicles under unknown-but-bounded noises. Then, a discrete reward based on thresholds of the trajectory tracking error is proposed for the reward function design, which guides the policy optimization algorithm to limit the tracking error within a certain range in order to improve the accuracy of trajectory tracking. Finally, simulation experiments demonstrate the effectiveness and superiority of the proposed method.

Source: Transactions of the Institute of Measurement and Control

DOI: 10.1177/01423312251388184

Authors: Jixu Zhao, Yilian Zhang, Zhenxiang Wang, Xianwen Zhou

Institution: Shanghai Maritime University

Language: English

Topic: Reinforcement Learning in Robotics

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: bronze

Funder: National Natural Science Foundation of China

Awards: G3583332013, G4806845379, G7019551738