Authors

Abstract

Orchard robots play a crucial role in agricultural production. Autonomous navigation serves as the foundation for orchard robots and eco-unmanned farms. Accurate sensing and localization are prerequisites for achieving autonomous navigation. However, current vision-based navigation solutions are sensitive to environmental factors, such as light, weather, and background, which can affect positioning accuracy. Therefore, they are unsuitable for outdoor navigation applications. LIDAR provides accurate distance measurements and is suitable for a wide range of environments. Its immunity to interference is not affected by light, colour, weather, or other factors, making it suitable for low objects and complex orchard scenes. Therefore, LiDAR navigation is more suitable for orchard environments. In complex orchard environments, tree branches and foliage can cause Global Positioning System (GNSS) accuracy to degrade, resulting in signal loss. Therefore, the major challenge that needs to be addressed is generating navigation paths and locating the position of orchard robots. In this paper, an improved method for Simultaneous Localization and Mapping (SLAM) and A-star algorithm is proposed. The SLAM and path planning method designed in this study effectively solves the problems of insufficient smoothness and large curvature fluctuation of the path planned in the complex orchard environment, and improves the detection efficiency of the robot. The experimental results indicate that the method can consistently and accurately fulfil the robot's detection needs in intricate orchard environments.

Abstract in Chinese

果园机器人在农业生产中发挥着至关重要的作用。自主导航是果园机器人和生态无人农场的基础。准确的感知 和定位是实现自主导航的先决条件。然而，目前基于视觉的导航解决方案对光线、天气和背景等环境因素非常 敏感，会影响定位精度。因此，它们不适合户外导航应用。激光雷达可提供精确的距离测量，适用于各种环境。它的抗干扰能力不受光线、颜色、天气或其他因素的影响，因此适用于低矮物体和复杂的果园场景。因此，激 光雷达导航更适合果园环境。在复杂的果园环境中，树枝和树叶会导致全球定位系统（GNSS）精度降低，造 成信号丢失。因此，需要解决的主要挑战是生成导航路径和定位果园机器人的位置。在本文中，我们提出了一种改进的同步定位与地图构建（SLAM）方法和 A-star 算法。本研究设计的 SLAM 和路径规划方法有效解决 了复杂果园环境下规划路径的平滑度不够和曲率波动较大的问题，提高了机器人的巡检效率。实验结果表明，该方法能稳定、准确地满足机器人在复杂果园环境中的巡检需求。