Scholarlink

<p>In complex environments, mobile robots need to consider various road factors and respond quickly to plan a reasonable path. Therefore, this paper proposes a hybrid adaptive genetic algorithm combining D*Lite and Simulated Annealing, termed D*Lite Simulated Annealing-Genetic Algorithm (DS-GA). DS-GA integrates the strengths of D*Lite, Simulated Annealing, and Genetic Algorithms. The optimized D*Lite algorithm is incorporated into the random initialization population step of the genetic algorithm, and an elite retention mechanism is introduced. The Simulated Annealing algorithm is employed to retain individuals with lower fitness values with a certain probability, optimizing the crossover and mutation operators as well as the fitness function. Through these strategies, the quality and diversity of the population are ensured while improving the optimal path, and the impact of multi-algorithm fusion on computational speed is minimized. Experimental results demonstrate that the improved DS-GA exhibits strong global search capabilities, enhanced local search performance, and excellent generalizability and robustness. Compared to other algorithms in the experiments, DS-GA achieves the shortest average optimal path.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Global path planning method based on improved genetic algorithm

  • 期刊论文

  • 发表日期：

    2026年1月6日

查看全文
  • Yuan Luo,
  • Jingxi Tan,
  • Yuheng Han

摘要

In complex environments, mobile robots need to consider various road factors and respond quickly to plan a reasonable path. Therefore, this paper proposes a hybrid adaptive genetic algorithm combining D*Lite and Simulated Annealing, termed D*Lite Simulated Annealing-Genetic Algorithm (DS-GA). DS-GA integrates the strengths of D*Lite, Simulated Annealing, and Genetic Algorithms. The optimized D*Lite algorithm is incorporated into the random initialization population step of the genetic algorithm, and an elite retention mechanism is introduced. The Simulated Annealing algorithm is employed to retain individuals with lower fitness values with a certain probability, optimizing the crossover and mutation operators as well as the fitness function. Through these strategies, the quality and diversity of the population are ensured while improving the optimal path, and the impact of multi-algorithm fusion on computational speed is minimized. Experimental results demonstrate that the improved DS-GA exhibits strong global search capabilities, enhanced local search performance, and excellent generalizability and robustness. Compared to other algorithms in the experiments, DS-GA achieves the shortest average optimal path.