Tree-SLAM: semantic object SLAM for efficient mapping of individual trees in orchards

2025-07-16

AI Summary

Weak GPS signals and high visual similarity among trees in orchards cause conventional SLAM systems to fail in mapping. To address this, this paper proposes a semantic SLAM framework using tree trunks as semantic anchors. The method jointly optimizes RGB-D visual features, noisy GPS measurements, and odometry data within a factor graph. It introduces an instance-segmentation-driven trunk detection module and designs a cascaded graph structure to enable cross-frame trunk re-identification and robust data association. Evaluated in real-world apple and pear orchards across multiple seasons, the system achieves a mean geolocation error of only 18 cm per individual tree—less than 20% of the typical inter-tree spacing. This demonstrates substantial improvements in mapping robustness and individual-tree localization accuracy within highly repetitive structural environments.

Technology Category

• Intelligent Robots: Localization, Mapping, and Navigation
• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Search and Optimization: Learning to Search

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding

Abstract

Accurate mapping of individual trees is an important component for precision agriculture in orchards, as it allows autonomous robots to perform tasks like targeted operations or individual tree monitoring. However, creating these maps is challenging because GPS signals are often unreliable under dense tree canopies. Furthermore, standard Simultaneous Localization and Mapping (SLAM) approaches struggle in orchards because the repetitive appearance of trees can confuse the system, leading to mapping errors. To address this, we introduce Tree-SLAM, a semantic SLAM approach tailored for creating maps of individual trees in orchards. Utilizing RGB-D images, our method detects tree trunks with an instance segmentation model, estimates their location and re-identifies them using a cascade-graph-based data association algorithm. These re-identified trunks serve as landmarks in a factor graph framework that integrates noisy GPS signals, odometry, and trunk observations. The system produces maps of individual trees with a geo-localization error as low as 18 cm, which is less than 20% of the planting distance. The proposed method was validated on diverse datasets from apple and pear orchards across different seasons, demonstrating high mapping accuracy and robustness in scenarios with unreliable GPS signals.

Problem

• Mapping individual trees in orchards with unreliable GPS signals
• Overcoming repetitive tree appearance challenges in standard SLAM
• Achieving high accuracy in geo-localization of orchard trees

Innovation

• Semantic SLAM for individual tree mapping
• Cascade-graph-based trunk re-identification algorithm
• Factor graph integrating GPS, odometry, and trunks