Energy-Aware PSO-Optimized Fuzzy Logic Control for Automated Guided Vehicles on Low-Power Edge Computing Platforms

Abstract

Automated Guided Vehicles (AGVs) serve as the backbone of modern smart factory logistics, yet their operational efficiency is often limited by sub optimal energy management during navigation. Conventional heuristic based Fuzzy Logic Controllers (FLCs) frequently exhibit oscillatory responses, leading to high current spikes and rapid battery degradation. This study proposes an energy aware Takagi-Sugeno-Kang (TSK) FLC, optimized using Particle Swarm Optimization (PSO) for deployment on a dual core ESP32 edge computing platform. A novel PSO fitness function was formulated to penalize both trajectory tracking errors and cumulative electrical power consumption, ensuring precise control with minimal computational overhead. Experimental validation on a 15-kg differential drive AGV demonstrated that the optimized controller achieved superior kinematic performance, with the Root Mean Square Error (RMSE) reduced by 68.48% (from 0.165 m to 0.052 m). Furthermore, the approach achieved a 16.84% reduction in average power consumption and effectively mitigated peak transient power spikes by 28.08%. These results demonstrate the viability of implementing advanced metaheuristic optimization on low-power, cost-effective microcontrollers. The proposed framework enhances operational sustainability, directly contributing to extended battery cycle life and reduced maintenance downtime in autonomous industrial environments.