Adaptive motion control for autonomous mobile robots: A comparative study of robust tracking under dynamic uncertainties

Abstract

• Provides a systematic evaluation of four controllers for differential-drive mobile robots under dynamic uncertainties: standard σ -adaptive, modified σ +PD adaptive, fuzzy self-tuning PD, and classical PD. • Proposes a σ +PD adaptive controller that embeds derivative terms inside the σ -law, achieving 38% reduction in angular Total Variation (TV) while maintaining robust tracking under 10–50% parametric variations. • Demonstrates that the fuzzy self-tuning PD controller attains the lowest tracking error (RMS 0.059 m), outperforming σ -adaptive and classical PD controllers in accuracy. • Employs MATLAB/Simulink simulations with a validated Pioneer 2-DX dynamic model to ensure reproducibility under realistic uncertainty and noise conditions. • Assesses performance using RMS error, TV of velocity commands, and Integral Absolute Error (IAE), ensuring a fair, multi-criteria evaluation across controllers. • Shows that the σ +PD adaptive controller offers the best trade-off between accuracy and smoothness, reducing chattering and lowering control effort. • Provides practical guidance for controller selection: fuzzy self-tuning is recommended when accuracy is critical, while σ +PD is preferable for stable, energy-efficient operation in uncertain environments. This study investigates motion control strategies for trajectory and position tracking of a unicycle-modeled mobile robot under dynamic parametric uncertainty. Four controllers are examined: an adaptive controller with a σ -modification term, a modified adaptive controller incorporating both σ and PD terms, a fuzzy self-tuning PD controller, and a classical PD controller. The evaluation considers two scenarios, circular path tracking with varying radii and point regulation from an initial pose. Controller performance is assessed using tracking error, Root Mean Square (RMS) error, Total Variation (TV) of linear and angular velocities, control effort, and velocity fluctuations, under both nominal and perturbed conditions. Results show that the modified adaptive controller reduces oscillations more effectively, whereas the fuzzy self-tuning controller achieves the lowest tracking errors. These outcomes provide practical guidance for selecting robust control strategies for differentially-driven robots operating in dynamic environments. All simulations were performed in MATLAB®/Simulink.