Scholarlink

This paper studies the visual navigation of vehicles in unknown environments. Due to the lack of global information, the accuracy of graphical navigation method will be decline. We propose a double LSTM attention navigation model to improve vehicle navigation performance. The model utilizes a dual-channel LSTM network for processing time series data and introduces an attention mechanism into the asynchronous dominant actor critic algorithm to mitigate the reward sparsity problem. Through comparative with various models on the AI2-THOR platform, the enhanced model improves the cumulative rewards and success rates of visual navigation.

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

Deep Reinforcement Learning Vehicle Visual Navigation Based on Attention Mechanism and LSTM

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Zhentao Yu,
  • Guojun Zhang,
  • Ruoting Li,
  • Peng Liu

摘要

This paper studies the visual navigation of vehicles in unknown environments. Due to the lack of global information, the accuracy of graphical navigation method will be decline. We propose a double LSTM attention navigation model to improve vehicle navigation performance. The model utilizes a dual-channel LSTM network for processing time series data and introduces an attention mechanism into the asynchronous dominant actor critic algorithm to mitigate the reward sparsity problem. Through comparative with various models on the AI2-THOR platform, the enhanced model improves the cumulative rewards and success rates of visual navigation.