Authors

Abstract

Aiming at autonomous weeding in narrow, complex orchards, this paper proposes an improved A* and DWA fusion path planning algorithm. Globally, a safety distance penalty in the A* cost function prevents paths from adhering to tree trunks. Locally, enhanced DWA modules, including dynamic obstacle angle thresholds, hierarchical lateral braking, and adaptive velocity window expansion, resolve trajectory oscillation and local minima. Validated in real orchards, the algorithm achieved a 0.029 m tracking deviation and over 95% dynamic obstacle avoidance success rate, providing robust and safe navigation for autonomous robots.

Abstract in Chinese

针对狭窄、复杂的果园自主除草场景，本文提出了一种改进的A与DWA融合路径规划算法。在全局规划中，A代价函数中的安全距离惩罚项可防止路径过分紧贴树干。在局部规划中，增强的DWA模块（包括动态障碍物角度阈值、分层侧向制动和自适应速度窗口扩展）解决了轨迹震荡和局部极小值问题。经真实果园环境验证，该算法的轨迹跟踪偏差为0.029米，动态避障成功率超过95%，为自主机器人提供了鲁棒且安全的导航保障。