A search and rescue robot search method based on flower pollination algorithm and Q-learning fusion algorithm.

Area of Science:

• Robotics
• Artificial Intelligence
• Path Planning

Background:

• Mobile robot motion planning is crucial for task completion.
• Complex environments pose significant challenges for traditional search algorithms.
• Existing algorithms often struggle with efficiency and adaptability.

Purpose of the Study:

• To propose a novel fusion algorithm combining Flower Pollination and Q-learning (FIQL) for enhanced robot path planning.
• To improve environment modeling using a hybrid static and dynamic grid map.
• To optimize Q-table initialization and reward functions for search and rescue scenarios.

Main Methods:

• Developed an improved grid map incorporating both static and dynamic elements.
• Integrated Q-learning with the Flower Pollination algorithm for Q-table initialization and path search.
• Implemented a combined static and dynamic reward function tailored to diverse environmental situations.

Main Results:

• The improved grid map significantly increased path planning success rates.
• The FIQL algorithm demonstrated superior performance in complex environments compared to other methods.
• FIQL reduced iteration counts, accelerated convergence, and required less computational effort.

Conclusions:

• The proposed FIQL algorithm effectively enables search and rescue robots to navigate complex environments.
• FIQL offers improved adaptability, efficiency, and reduced computational load for mobile robot path planning.
• The hybrid grid map and reward function contribute to enhanced performance and success rates.