Scholarlink

In off-road scenarios, the combined impact of obstacles and terrain poses a significant challenge to vehicle navigation. Currently, many studies focus on static environments, and these methods fail to adapt to dynamic environments due to their neglect of moving obstacles. Therefore, based on the Q Learning algorithm, this paper proposes a method for initializing the Q-table using a global route and designs elevation difference and distance reward functions to enable effective path planning in low-dynamic off-road environments. Experimental results demonstrate that compared to an untuned Q Learning algorithm, this method converges faster and produces smoother routes, thereby better adapting to dynamic off-road environments.

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

Intelligent Path Planning for Off-Road Navigation: A Q Learning Approach with Terrain and Distance Optimization

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Duohang Sun,
  • Chenxi Wang,
  • Chengxu Zhou,
  • Xiaoqing Liu,
  • Haitao Xiao

摘要

In off-road scenarios, the combined impact of obstacles and terrain poses a significant challenge to vehicle navigation. Currently, many studies focus on static environments, and these methods fail to adapt to dynamic environments due to their neglect of moving obstacles. Therefore, based on the Q Learning algorithm, this paper proposes a method for initializing the Q-table using a global route and designs elevation difference and distance reward functions to enable effective path planning in low-dynamic off-road environments. Experimental results demonstrate that compared to an untuned Q Learning algorithm, this method converges faster and produces smoother routes, thereby better adapting to dynamic off-road environments.