Scholarlink

<p>With the continuous advancement of technology and industrial automation, the application of robotic arms in complex high-dimensional environments is becoming increasingly prevalent. Trajectory planning is a key issue in robotics, especially in complex obstacle environments. To address this challenge, this study proposes a deep reinforcement learning algorithm based on the combination of long short-term memory networks and the artificial potential field method, named LSTM–APF–DDQN, designed to improve the trajectory planning efficiency and obstacle avoidance capabilities of robotic arms in complex environments. To ensure the practical applicability of the algorithm, joint simulation validation using Python and Unity was carried out. The results demonstrate that the proposed algorithm achieves high effectiveness and reliability in complex high-dimensional environments. This algorithm provides strong technical support for future robotic arm trajectory planning in complex environments and has broad application prospects.</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

Intelligent robotic arm trajectory planning using improved classical Q-learning and LSTM

  • 期刊论文

  • 发表日期：

    2026年1月5日

查看全文
  • Jiaqiang Zhang,
  • Huiqing Chen,
  • Yixiong Wang,
  • Hongzhen Xu,
  • Tianxiang Yan

摘要

With the continuous advancement of technology and industrial automation, the application of robotic arms in complex high-dimensional environments is becoming increasingly prevalent. Trajectory planning is a key issue in robotics, especially in complex obstacle environments. To address this challenge, this study proposes a deep reinforcement learning algorithm based on the combination of long short-term memory networks and the artificial potential field method, named LSTM–APF–DDQN, designed to improve the trajectory planning efficiency and obstacle avoidance capabilities of robotic arms in complex environments. To ensure the practical applicability of the algorithm, joint simulation validation using Python and Unity was carried out. The results demonstrate that the proposed algorithm achieves high effectiveness and reliability in complex high-dimensional environments. This algorithm provides strong technical support for future robotic arm trajectory planning in complex environments and has broad application prospects.