Comparison of Bioinspired Tuning of PID Controllers for Mobile Robots Obstacle Following/Avoidance

M. A. Pastrana; J. Bautista; Jose Mendoza-Peñaloza; L. H. Oliveira; William Humberto Cuéllar Sánchez; Daniel M. Muñoz

doi:10.20906/SBAI-SBSE-2023/3951

M. A. Pastrana Department of Mechanical Engineering, Mechatronics Graduate Program, University of Brasilia, Brasilia, DF
J. Bautista University of the Amazonia, Faculty of Engineering, Systems Undergraduate Program, Florencia, Caquetá
Jose Mendoza-Peñaloza Department of Mechanical Engineering, Mechatronics Graduate Program, University of Brasilia, Brasilia, DF
L. H. Oliveira Department of Mechanical Engineering, Mechatronics Graduate Program, University of Brasilia, Brasilia, DF
William Humberto Cuéllar Sánchez Department of Mechanical Engineering, Mechatronics Graduate Program, University of Brasilia, Brasilia, DF
Daniel M. Muñoz Faculty of Gama, Electronics Engineering Undergraduate Program, University of Brasilia, Brasilia, DF / Department of Mechanical Engineering, Mechatronics Graduate Program, University of Brasilia, Brasilia, DF

DOI: https://doi.org/10.20906/SBAI-SBSE-2023/3951

Keywords: Control systems, PID controller, Bioinspired algorithms, mobile robots, test hypotheses

Abstract

Currently, Proportional Integral and Derivative (PID) controllers are widely used in educational and industrial settings due to their robustness and ease of implementation. However, classical tuning methods like Ziegler-Nichols or Root Locus do not guarantee optimal tuning for the PID controller and require previous skills in control systems. Bioinspired optimization strategies can be employed to obtain suboptimal solutions for specific applications, without requiring extensive knowledge of control system design. In this study, a PID controller applied to mobile robot obstacle following/avoidance was tuned using four bioinspired algorithms, namely Particle Swarm Optimization (PSO), Differential Evolution (DE), Grasshopper Optimization Algorithm (GOA), and Moth Flame Optimization (MFO). The effectiveness of these algorithms was evaluated on a single-input single-output mobile robot platform through six different scenarios. Furthermore, statistical tests were conducted to determine the best-tuned bioinspired PID controller based on criteria such as overshoot, settling time, and steady-state error. Results showed that the MFO-PID was the best controller in five scenarios. Furthermore, a comparison between MFO-PID and classical PID was conducted to determine the best PID controller. The MFO-PID showed better success rates with a shorter settling time in all scenarios (100%), a lower steady-state error in four scenarios (66.67%), and less overshot in three scenarios (50%). Finally, the MFO-PID was the only one that stabilized in the most challenging scenarios.