**OrchLoc: In-Orchard Localization via a Single LoRa Gateway and Generative Diffusion Model-based Fingerprinting** **Authors:** Kang Yang **Abstract:** In orchards, tree-level localization of robots is crucial for smart agriculture applications such as precision disease management and targeted nutrient dispensing. However, existing solutions often fail to provide adequate accuracy. This paper introduces OrchLoc, a novel fingerprinting-based localization system that achieves tree-level accuracy with only one LoRa gateway. The system extracts Channel State Information (CSI) from eight channels to form the fingerprint. To avoid labor-intensive site surveys, a Generative Model (CGM) is designed to learn the relationship between CSIs and their corresponding locations. The CGM is fine-tuned using CSIs from static LoRa sensor nodes to build and update the fingerprint database. Extensive experiments in two pistachio orchards validate the system's effectiveness, achieving tree-level localization with minimal overhead and enhancing robot navigation accuracy. **Key Contributions:** 1. **OrchLoc System:** A fingerprinting-based localization system that achieves tree-level accuracy with a single LoRa gateway. 2. **CSI-based Fingerprint:** Extracts CSI from eight channels to enhance localization accuracy. 3. **CGM Model:** A generative model that learns the relationship between CSIs and locations, enabling efficient database construction and updates. 4. **Turbo-Training Scheme:** Combines pre-training and fine-tuning to maintain high accuracy over time. 5. **Performance Evaluation:** Achieves average precision and recall of 96.3% and 97.6%, respectively, over four weeks, with a localization error of 1.2 m in ten areas. **Introduction:** In modern orchards, robots are essential for precision agriculture practices. However, traditional localization methods like wheel encoders, SLAM, and GPS/INS systems face significant challenges in achieving tree-level accuracy in orchards. This paper addresses these issues by leveraging existing LoRa infrastructure to enable robot localization. **Methodology:** 1. **CSI Extraction:** Extracts CSI from eight channels to form the fingerprint. 2. **CGM Model:** A generative model that learns the relationship between CSIs and locations, using a location-aware diffusion model (LoDM) and CSI and location representers. 3. **Turbo-Training Scheme:** Combines pre-training and fine-tuning to maintain high accuracy over time. 4. **Experiments:** Conducted in two pistachio orchards, demonstrating the system's effectiveness in achieving tree-level localization with minimal overhead. **Results:** - **Pistachio Orchard 1:** Achieves average precision and recall of 96.3% and 97.6%, respectively, over four weeks. - **Pistachio Orchard 2:** Achieves a localization error of 1.2 m in ten areas. **Conclusion:** OrchLoc provides a novel**OrchLoc: In-Orchard Localization via a Single LoRa Gateway and Generative Diffusion Model-based Fingerprinting** **Authors:** Kang Yang **Abstract:** In orchards, tree-level localization of robots is crucial for smart agriculture applications such as precision disease management and targeted nutrient dispensing. However, existing solutions often fail to provide adequate accuracy. This paper introduces OrchLoc, a novel fingerprinting-based localization system that achieves tree-level accuracy with only one LoRa gateway. The system extracts Channel State Information (CSI) from eight channels to form the fingerprint. To avoid labor-intensive site surveys, a Generative Model (CGM) is designed to learn the relationship between CSIs and their corresponding locations. The CGM is fine-tuned using CSIs from static LoRa sensor nodes to build and update the fingerprint database. Extensive experiments in two pistachio orchards validate the system's effectiveness, achieving tree-level localization with minimal overhead and enhancing robot navigation accuracy. **Key Contributions:** 1. **OrchLoc System:** A fingerprinting-based localization system that achieves tree-level accuracy with a single LoRa gateway. 2. **CSI-based Fingerprint:** Extracts CSI from eight channels to enhance localization accuracy. 3. **CGM Model:** A generative model that learns the relationship between CSIs and locations, enabling efficient database construction and updates. 4. **Turbo-Training Scheme:** Combines pre-training and fine-tuning to maintain high accuracy over time. 5. **Performance Evaluation:** Achieves average precision and recall of 96.3% and 97.6%, respectively, over four weeks, with a localization error of 1.2 m in ten areas. **Introduction:** In modern orchards, robots are essential for precision agriculture practices. However, traditional localization methods like wheel encoders, SLAM, and GPS/INS systems face significant challenges in achieving tree-level accuracy in orchards. This paper addresses these issues by leveraging existing LoRa infrastructure to enable robot localization. **Methodology:** 1. **CSI Extraction:** Extracts CSI from eight channels to form the fingerprint. 2. **CGM Model:** A generative model that learns the relationship between CSIs and locations, using a location-aware diffusion model (LoDM) and CSI and location representers. 3. **Turbo-Training Scheme:** Combines pre-training and fine-tuning to maintain high accuracy over time. 4. **Experiments:** Conducted in two pistachio orchards, demonstrating the system's effectiveness in achieving tree-level localization with minimal overhead. **Results:** - **Pistachio Orchard 1:** Achieves average precision and recall of 96.3% and 97.6%, respectively, over four weeks. - **Pistachio Orchard 2:** Achieves a localization error of 1.2 m in ten areas. **Conclusion:** OrchLoc provides a novel