Fruit Tree Row Recognition and 2D Map Generation for Autonomous Driving in Orchards

We present a novel algorithm for creating 2D maps tailored for autonomous navigation within orchards. Recognizing that fruit trees in orchards are typically aligned in rows, our primary goal is to accurately detect these tree rows and project this information onto the map. Initially, we propose a simple algorithm that recognizes trees from point cloud data by analyzing the spatial distribution of points. We then introduce a method for detecting fruit tree rows based on the positions of recognized fruit trees, which are integrated into the 2D orchard map. Validation of the proposed approach was conducted using real-world orchard point cloud data acquired via LiDAR. The results demonstrate high tree detection accuracy of 90% and precise tree row mapping, confirming the method