Abstract: In the logistics industry, while unmanned forklifts is gradually being applied. A very important feature of unmanned forklifts is the ability to detect the position of the pallet holes used for loading, so that the forklift can insert the pallets to transport the cargo. Machine learning algorithms are widely used to automatically detect the position of an object, such as the location of a hole. This paper describes the development of a deep learning approach to identifying the position of pallet holes using YOLO and Raspberry Pi. The model uses the YOLOv5 model, and the model is trained with a self-built pallet dataset, and the algorithm is deployed into an embedded device to achieve pallet hole recognition. The experimental results show that the recognition accuracy of pallet hole is around 88% while meeting the real-time performance requirements, which is low cost and high speed compared to the main methods currently used in the industry.

Key words: Unmanned forklift; Machine learning; YOLO; Raspberry Pi; Forklift pallet

摘要： 在物流领域中，无人驾驶叉车逐渐被广泛应用。无人驾驶叉车需要具备一项非常重要 的功能，即可检测到装载货物用的托盘孔的位置，从而使叉车插入托盘来运输货物，其中机 器学习算法被广泛用于自动检测物体的位置领域，如孔洞的位置检测。文章介绍了使用 YOLO 和树莓派开发深度学习识别托盘孔位置的方法，采用 YOLOv5 模型，通过自建的托盘数据集 对模型进行训练，并将算法部署到嵌入式设备中，实现了托盘孔识别。实验结果表明，在满 足实时性能的要求下，对托盘孔识别准确率为 88%左右，与当前行业主要使用的方法相比， 具有低成本高速度的特点。

关键词: 无人驾驶叉车；机器学习；YOLO；树莓派；叉车托盘