Scholarlink

<p>To address issues such as insufficient initial exploration, susceptibility to local optima, and lack of dynamic obstacle avoidance capabilities in traditional path planning, this paper proposes a two-layer hybrid algorithm for dynamic obstacle avoidance path planning. The upper layer employs an improved gray wolf optimization algorithm to plan globally optimal paths and extract key nodes. In contrast, the lower layer utilizes these nodes as guides to perform real-time local dynamic obstacle avoidance through an enhanced dynamic window method. To optimize the Gray Wolf algorithm, we introduced an improved sine mapping with an offset term and a dynamic perturbation term to initialize the population, significantly enhancing population diversity. Simultaneously, its position update strategy incorporates proportional weighting based on step-length Euclidean distance and draws inspiration from the concept of retaining optimal historical positions in particle swarm optimization. This effectively enhances the algorithm’s performance during both the global exploration and local search phases. For the dynamic window method, this paper designed a dynamic prediction time domain based on environmental complexity. Additionally, a novel evaluation function has been developed. Building upon the traditional three metrics, it incorporates new components: a dynamic obstacle trend indicator and a global path alignment indicator. This enables more precise control over obstacle avoidance safety, reducing collisions and local deadlocks. Simulation and experimental results demonstrate that the proposed hybrid algorithm achieves significant improvements in both dynamic obstacle avoidance and path planning quality compared to existing algorithms.</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

Adaptive trajectory planning for logistics vehicles: integrating multi-strategy gray wolf optimization and enhanced dynamic window approach

  • 期刊论文

  • 发表日期：

    2026年5月23日

查看全文
  • Shaoxiong Shi,
  • Weilin Wu,
  • Zhongjian Xie,
  • Yujie Lin

摘要

To address issues such as insufficient initial exploration, susceptibility to local optima, and lack of dynamic obstacle avoidance capabilities in traditional path planning, this paper proposes a two-layer hybrid algorithm for dynamic obstacle avoidance path planning. The upper layer employs an improved gray wolf optimization algorithm to plan globally optimal paths and extract key nodes. In contrast, the lower layer utilizes these nodes as guides to perform real-time local dynamic obstacle avoidance through an enhanced dynamic window method. To optimize the Gray Wolf algorithm, we introduced an improved sine mapping with an offset term and a dynamic perturbation term to initialize the population, significantly enhancing population diversity. Simultaneously, its position update strategy incorporates proportional weighting based on step-length Euclidean distance and draws inspiration from the concept of retaining optimal historical positions in particle swarm optimization. This effectively enhances the algorithm’s performance during both the global exploration and local search phases. For the dynamic window method, this paper designed a dynamic prediction time domain based on environmental complexity. Additionally, a novel evaluation function has been developed. Building upon the traditional three metrics, it incorporates new components: a dynamic obstacle trend indicator and a global path alignment indicator. This enables more precise control over obstacle avoidance safety, reducing collisions and local deadlocks. Simulation and experimental results demonstrate that the proposed hybrid algorithm achieves significant improvements in both dynamic obstacle avoidance and path planning quality compared to existing algorithms.