QL-HIT2F: A Q-Learning-Aided Adaptive Fuzzy Path Planning Algorithm with Enhanced Obstacle Avoidance

Area of Science:

• Robotics
• Artificial Intelligence
• Control Systems

Background:

• Fuzzy logic methods are widely used for robot path planning.
• Genetic Algorithm-based Hierarchical Interval Type-2 Fuzzy (GA-HIT2F) systems offer novel planning capabilities.
• Existing GA-HIT2F methods exhibit limitations in adaptability and flexibility, hindering application in complex scenarios and leading to planning failures.

Purpose of the Study:

• To address the limitations of the GA-HIT2F approach for robot path planning.
• To propose an enhanced Q-Learning-aided Adaptive Hierarchical Interval Type-2 Fuzzy (QL-HIT2F) algorithm.
• To improve robot collision avoidance with special obstacles and optimize angular adjustments.

Main Methods:

• Developed a Q-Learning-aided Adaptive Hierarchical Interval Type-2 Fuzzy (QL-HIT2F) algorithm.
• Incorporated reinforcement learning for enhanced collision avoidance.
• Introduced average obstacle orientation (AOO) for optimizing angular adjustments.
• Integrated supplementary robot parameters and fuzzy membership parameters into the reinforcement learning action space.
• Utilized meta-map and sub-training concepts during the training process.

Main Results:

• The proposed QL-HIT2F algorithm demonstrates improved adaptability and flexibility in path planning.
• Enhanced capability in avoiding collisions with special obstacles through reinforcement learning.
• Optimized robot angular adjustments using the average obstacle orientation (AOO).
• Simulation results validate the feasibility and effectiveness of the QL-HIT2F approach.

Conclusions:

• The QL-HIT2F algorithm represents a significant advancement in robot path planning.
• The integration of Q-learning and fuzzy logic enhances navigation in complex environments.
• The proposed method offers a more robust and adaptable solution compared to previous GA-HIT2F systems.