机电工程技术2025，Vol.54Issue(24)：49-53,5.DOI:10.3969/j.issn.1009-9492.2025.24.009
一种自学习魔方还原算法的设计与实现
Design and Implementation of a Self-learning Rubik's Cube Solving Algorithm
孔凡国 1袁功兴 1陈靖轩1
作者信息
- 1. 五邑大学机械与自动化工程学院,广东 江门 529020
- 折叠
摘要
Abstract
With the rapid development of artificial intelligence,the application of reinforcement learning in various games has become one of the research hotspots.One goal of reinforcement learning is to develop algorithms that can learn and surpass human proficiency in challenging domains.A self-learning Rubik's Cube solving algorithm is designed based on Monte Carlo Tree Search(MCTS)and deep neural networks,which starts from scratch without human knowledge to solve scrambled Rubik's Cubes.The study employs a deep reinforcement learning method based on the DQN algorithm,using MCTS to predict the optimal move by simulating the outcomes of different action paths.By combining Rubik's Cube solving with MCTS and deep neural networks,a trained neural network model is introduced into the MCTS process to replace the actual search results with neural network predictions for state evaluation.Through MCTS,the algorithm solves Rubik's Cube by training the neural network to find the optimal solution path.Simulation results show that the algorithm achieves a 90%success rate in solving Rubik's Cube,with an average of 21 steps to complete the solution.This demonstrates the feasibility of the method.关键词
强化学习/启发式搜索/魔方还原/深度神经网络/人工智能Key words
reinforcement learning/heuristic search/Rubik's Cube solving/deep neural network/artificial intelligence分类
信息技术与安全科学引用本文复制引用
孔凡国,袁功兴,陈靖轩..一种自学习魔方还原算法的设计与实现[J].机电工程技术,2025,54(24):49-53,5.
