Watering ornamental plants is considered to be a time-consuming task with the increasing garden size. Employing robots to conduct the watering task is a feasible way to improve the watering efficiency. However, general path planning models for watering robots are faced with various difficulties, such as the curse of dimensionality caused by the excessively oversized irrigation area and insufficient irrigation due to weather conditions. To tackle these issues, this paper proposes two novel strategies for watering robot optimal path modeling. Firstly, for tackling the curse of dimensionality, a sliding window strategy is proposed. To be specific, the whole watering area is partitioned into many subareas using the sliding window, and each subarea is watered independently. Secondly, to overcome the problem that the soil moisture still can not reach the expectation due to various weather conditions, this paper proposes a placeholder strategy, which enables the watering robot to dynamically adjust the watering path. To test these strategies, a novel genetic algorithm with neighbor exchanging strategy is proposed. Extensive experiments demonstrate the effectiveness of watering robot path planning models based on the proposed strategies.

随着花园面积的增加，给观赏植物浇水被认为是一项耗时的工作。用机器人来执行浇水任务是提高浇水效率的一种可行方法。然而，用于浇水机器人的一般路径规划模型面临各种困难，例如由于灌溉面积过大和天气条件导致灌溉不足而引起的尺寸诅咒。为了解决这些问题，本文提出了两种新颖的浇水机器人最佳路径建模策略。首先，为解决维数问题，提出了一种滑动窗口策略。具体而言，使用滑动窗将整个浇水区域划分为多个子区域，并且每个子区域被独立地浇水。第二，为克服各种天气条件下土壤水分仍达不到预期的问题，提出一种占位策略，使浇水机器人能够动态调整浇水路径。为了测试这些策略，提出了一种具有邻居交换策略的新型遗传算法。大量实验证明了基于所提出策略的浇水机器人路径规划模型的有效性。