Predictive PID Control for Automated Guided Vehicles Using Genetic Algorithm and Machine Learning

Predictive PID Control for Automated Guided Vehicles Using Genetic Algorithm and Machine Learning

The integration of automated guided vehicles (AGVs) in industrial automation demands precise and adaptive control systems for efficient path tracking. This study introduces a hybrid framework combining traditional Proportional-Integral-Derivative (PID) control with advanced machine learning to optim...

Full description

Saved in:
Bibliographic Details
Main Authors: Kinza Nazir, Yong-Woon Kim, Yung-Cheol Byun
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Automated guided vehicle
genetic algorithm
machine learning
path tracking
support vector regression
Online Access:https://ieeexplore.ieee.org/document/10960296/
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The integration of automated guided vehicles (AGVs) in industrial automation demands precise and adaptive control systems for efficient path tracking. This study introduces a hybrid framework combining traditional Proportional-Integral-Derivative (PID) control with advanced machine learning to optimize AGV performance. A genetic algorithm (GA) was employed to generate ground truth PID parameters for diverse track configurations, ensuring superior path-tracking accuracy. However, the computational intensity of GA inspired the adoption of Support Vector Regression (SVR) as a scalable and efficient alternative for predicting PID gains. A custom simulator, designed with Webots, facilitated realistic testing environments, and comprehensive dataset with systematically varied track geometries was curated for training and evaluation. Results demonstrate the proposed framework’s ability to maintain high tracking accuracy while significantly reducing computational costs, with SVR predictions closely approximating the GA-derived benchmarks. Specifically, the GA-optimized true PID values achieved a mean error of 11.8402, while the SVR-predicted PID values had a mean error of 15.4438, with SVR attaining a recall of 86.55%. This synergy between heuristic optimization and machine learning advances the scalability and real-time adaptability of AGV control systems, paving the way for more robust industrial automation solutions.
ISSN:2169-3536

Similar Items