Scholarlink

<p>The global demand for agricultural products is on the rise, prompting an increase in the utilization of robots to automate harvesting tasks due to challenges such as aging farmers, labor shortages, and urbanization. In various scenarios, crops can be grasped in multiple directions, providing flexibility in path planning. The planner does not need to target a single configuration, as multiple configurations can represent the same goal position in the task space. This study introduces the construction of multi-goal rapidly exploring random trees star connect (multi-goal RRT*-Connect), employing a forward tree with the initial configuration and a backward tree containing a set of goal candidates. Since achieving a fast yet high-quality path is essential for applications such as harvesting tasks, we implemented a biased sampling method called local gap sampling to restrict the sampling space between bounded paths. The performance of the proposed multi-goal RRT*-Connect and local gap sampling was evaluated in simulation environments with a 2-degree-of-freedom (2-DOF) planar robot, a 3-DOF planar robot, and a 6-DOF robot manipulator, and through laboratory experiments.</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

Robot Path Planning With Grasping Pose Flexibility Incorporating Local Gap Sampling Approach

  • 期刊论文

  • 发表日期：

    2025年6月11日

查看全文
  • Phayuth Yonrith,
  • Ayoung Hong

摘要

The global demand for agricultural products is on the rise, prompting an increase in the utilization of robots to automate harvesting tasks due to challenges such as aging farmers, labor shortages, and urbanization. In various scenarios, crops can be grasped in multiple directions, providing flexibility in path planning. The planner does not need to target a single configuration, as multiple configurations can represent the same goal position in the task space. This study introduces the construction of multi-goal rapidly exploring random trees star connect (multi-goal RRT*-Connect), employing a forward tree with the initial configuration and a backward tree containing a set of goal candidates. Since achieving a fast yet high-quality path is essential for applications such as harvesting tasks, we implemented a biased sampling method called local gap sampling to restrict the sampling space between bounded paths. The performance of the proposed multi-goal RRT*-Connect and local gap sampling was evaluated in simulation environments with a 2-degree-of-freedom (2-DOF) planar robot, a 3-DOF planar robot, and a 6-DOF robot manipulator, and through laboratory experiments.