An Improved Q-Learning Algorithm for Path Planning

To address the issues of slow convergence speed and poor path planning performance in dynamic obstacle environments. This paper proposes an improved Q-Learning path planning algorithm for mobile robots. The algorithm introduces an exploration factor based on the mutation of probability to balance exploration and utilization and accelerate learning efficiency. Additionally, deep learning factors are designed in the update function to ensure the algorithm explores with probability. The integration of the Blending Inheritance algorithm prevents local path optimization and explores the optimal iteration step number by stages, reducing the repetition rate of dynamic map exploration.We conducted an experiment using the grid method to construct a map. Comparative experimental results showed significant optimization in terms of iteration times and path compared to traditional algorithms. The improved algorithm can better achieve path planning under dynamic maps, further confirming the practicality and effectiveness of the proposed method.