Adaptive fuzzy-learning control with multi-objective optimization for robust mobile robot navigation

This paper presents an integrated adaptive mobile robot navigation framework, the Intelligent Obstacle Dynamics and Path Complexity System (IODPCS), which integrates dual-layer fuzzy logic, artificial potential fields, reinforcement learning, and human-in-the-loop tuning into a unified adaptive control architecture. Unlike conventional fuzzy–potential field methods, IODPCS employs a two-level fuzzy reasoning structure: a global layer for terrain-aware speed and safety management, and a local layer for dynamic obstacle avoidance, both continuously adapted through real-time learning. A terrain-adaptive speed optimization strategy, combined with an actuator torque–velocity energy model, enables proactive trade-offs between traversal speed and efficiency. Operator demonstrations initialize structured rules, while reinforcement learning ensures long-term self-improvement. MATLAB/Simulink and hardware-in-the-loop experiments demonstrate that IODPCS achieves up to 35% shorter path lengths, 28% lower energy consumption, and collision-free navigation in both static and dynamic obstacle fields. These results demonstrate that IODPCS provides a robust framework for autonomous navigation in complex environments.