Scholarlink

<p>Autonomous vehicles (AVs) operating in off-road environments, for instance, in rescue missions and material transportation, must rapidly identify efficient and safe paths through complex terrains. Traditional rapidly exploring random tree (RRT) algorithms often exhibit slow convergence and inadequately consider diverse terrain features, resulting in suboptimal or infeasible paths. To address these limitations, this paper introduces the Multi-terrain RRT* (MT-RRT*), a novel path planning algorithm specifically optimized for complex off-road scenarios. MT-RRT* integrates multiple terrain obstacle layers, enabling differentiation between passable and impassable areas based on vehicle capabilities. By employing a biased sampling strategy and optimized parent selection, the algorithm achieves faster convergence.&#xa0;MT-RRT* calculates path costs by considering vehicle passability across diverse terrains using customizable cost functions with user-defined parameters. This approach facilitates the generation of optimal paths tailored to specific mission requirements and vehicle types, thereby enhancing both operational efficiency and safety. Simulation results demonstrate that MT-RRT* significantly improves path quality, reducing path elevation changes by over 31% compared to conventional RRT* algorithms, with only a minimal increase in overall path cost. Moreover, the proposed algorithm consistently identifies cost-effective paths even under constrained sampling durations, establishing it as a robust and adaptable solution for off-road AV navigation.</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

Multi-terrain RRT*: Terrain Aware Path Planning for Off-Road Environments

  • 期刊论文

  • 发表日期：

    2026年4月2日

查看全文
  • Qianlong Peng,
  • Jin Zhao,
  • Qingwen Meng,
  • Chaoyi Chen,
  • Qing Xu,
  • Jianqiang Wang,
  • Guangwei Wang

摘要

Autonomous vehicles (AVs) operating in off-road environments, for instance, in rescue missions and material transportation, must rapidly identify efficient and safe paths through complex terrains. Traditional rapidly exploring random tree (RRT) algorithms often exhibit slow convergence and inadequately consider diverse terrain features, resulting in suboptimal or infeasible paths. To address these limitations, this paper introduces the Multi-terrain RRT* (MT-RRT*), a novel path planning algorithm specifically optimized for complex off-road scenarios. MT-RRT* integrates multiple terrain obstacle layers, enabling differentiation between passable and impassable areas based on vehicle capabilities. By employing a biased sampling strategy and optimized parent selection, the algorithm achieves faster convergence. MT-RRT* calculates path costs by considering vehicle passability across diverse terrains using customizable cost functions with user-defined parameters. This approach facilitates the generation of optimal paths tailored to specific mission requirements and vehicle types, thereby enhancing both operational efficiency and safety. Simulation results demonstrate that MT-RRT* significantly improves path quality, reducing path elevation changes by over 31% compared to conventional RRT* algorithms, with only a minimal increase in overall path cost. Moreover, the proposed algorithm consistently identifies cost-effective paths even under constrained sampling durations, establishing it as a robust and adaptable solution for off-road AV navigation.