2026
1.
Deanna Flynn; Abhinav Jain; Heather Knight; Cristina G. Wilson; Cindy Grimm
Uncovering Implementable Dormant Pruning Decisions from Three Different Stakeholder Perspectives Journal Article
In: 2026.
Abstract | Links | BibTeX | Tags: AI, Labor
@article{flynn_uncovering_2026,
title = {Uncovering Implementable Dormant Pruning Decisions from Three Different Stakeholder Perspectives},
author = {Deanna Flynn and Abhinav Jain and Heather Knight and Cristina G. Wilson and Cindy Grimm},
url = {https://journals.ashs.org/view/journals/horttech/36/2/article-p325.xml},
doi = {10.21273/HORTTECH05817-25},
year = {2026},
date = {2026-04-01},
urldate = {2026-04-01},
chapter = {HortTechnology},
abstract = {Uncovering Implementable Dormant Pruning Decisions from Three Different Stakeholder Perspectives},
keywords = {AI, Labor},
pubstate = {published},
tppubtype = {article}
}
Uncovering Implementable Dormant Pruning Decisions from Three Different Stakeholder Perspectives
2.
Konstantinos I. Roumeliotis; Ranjan Sapkota; Manoj Karkee; Nikolaos D. Tselikas
Agentic AI With Orchestrator-Agent Trust: A Modular Visual Classification Framework With Trust-Aware Orchestration and RAG-Based Reasoning Journal Article
In: IEEE Access, vol. 14, pp. 26965–26982, 2026, ISSN: 2169-3536.
Abstract | Links | BibTeX | Tags: AI, Labor
@article{roumeliotis_agentic_2026b,
title = {Agentic AI With Orchestrator-Agent Trust: A Modular Visual Classification Framework With Trust-Aware Orchestration and RAG-Based Reasoning},
author = {Konstantinos I. Roumeliotis and Ranjan Sapkota and Manoj Karkee and Nikolaos D. Tselikas},
url = {https://ieeexplore.ieee.org/document/11373381/},
doi = {10.1109/ACCESS.2026.3662282},
issn = {2169-3536},
year = {2026},
date = {2026-01-01},
urldate = {2026-01-01},
journal = {IEEE Access},
volume = {14},
pages = {26965\textendash26982},
abstract = {Modern Artificial Intelligence (AI) increasingly relies on multi-agent architectures that blend visual and language understanding. Yet, a pressing challenge remains: How can we trust these agents especially in zero-shot settings with no fine-tuning? We introduce a novel modular Agentic AI visual classification framework that integrates generalist multimodal agents with a non-visual reasoning orchestrator and a Retrieval-Augmented Generation (RAG) module. Applied to apple leaf disease diagnosis, we benchmark three configurations: (I) zero-shot with confidence-based orchestration, (II) fine-tuned agents with improved performance, and (III) trust-calibrated orchestration enhanced by CLIP-based image retrieval and re-evaluation loops. Using confidence calibration metrics (ECE, OCR, CCC), the orchestrator modulates trust across agents. Our results demonstrate a 77.94% accuracy improvement in the zero-shot setting using trust-aware orchestration and RAG, achieving 85.63% overall. GPT-4o showed better calibration, while Qwen-2.5-VL displayed overconfidence. Furthermore, image-RAG grounded predictions with visually similar cases, enabling correction of agent overconfidence via iterative re-evaluation. The proposed system separates perception (vision agents) from meta-reasoning (orchestrator), enabling scalable and interpretable multi-agent AI. This blueprint illustrates how Agentic AI can deliver trustworthy, modular, and transparent reasoning, and is extensible to diagnostics, biology, and other trust-critical domains. In doing so, we highlight Agentic AI not just as an architecture but as a paradigm for building reliable multi-agent intelligence. All models, prompts, results, and system components including the complete software source code are openly released to support reproducibility, transparency, and community benchmarking at our Github page.},
keywords = {AI, Labor},
pubstate = {published},
tppubtype = {article}
}
Modern Artificial Intelligence (AI) increasingly relies on multi-agent architectures that blend visual and language understanding. Yet, a pressing challenge remains: How can we trust these agents especially in zero-shot settings with no fine-tuning? We introduce a novel modular Agentic AI visual classification framework that integrates generalist multimodal agents with a non-visual reasoning orchestrator and a Retrieval-Augmented Generation (RAG) module. Applied to apple leaf disease diagnosis, we benchmark three configurations: (I) zero-shot with confidence-based orchestration, (II) fine-tuned agents with improved performance, and (III) trust-calibrated orchestration enhanced by CLIP-based image retrieval and re-evaluation loops. Using confidence calibration metrics (ECE, OCR, CCC), the orchestrator modulates trust across agents. Our results demonstrate a 77.94% accuracy improvement in the zero-shot setting using trust-aware orchestration and RAG, achieving 85.63% overall. GPT-4o showed better calibration, while Qwen-2.5-VL displayed overconfidence. Furthermore, image-RAG grounded predictions with visually similar cases, enabling correction of agent overconfidence via iterative re-evaluation. The proposed system separates perception (vision agents) from meta-reasoning (orchestrator), enabling scalable and interpretable multi-agent AI. This blueprint illustrates how Agentic AI can deliver trustworthy, modular, and transparent reasoning, and is extensible to diagnostics, biology, and other trust-critical domains. In doing so, we highlight Agentic AI not just as an architecture but as a paradigm for building reliable multi-agent intelligence. All models, prompts, results, and system components including the complete software source code are openly released to support reproducibility, transparency, and community benchmarking at our Github page.
2025
3.
Miranda Cravetz; Purva Vyas; Cindy Grimm; Joseph R. Davidson
Slip detection for compliant robotic hands using inertial signals and deep learning Journal Article
In: Frontiers in Robotics and AI, vol. 12, 2025, ISSN: 2296-9144.
Abstract | Links | BibTeX | Tags: AI, Labor
@article{cravetz_slip_2025b,
title = {Slip detection for compliant robotic hands using inertial signals and deep learning},
author = {Miranda Cravetz and Purva Vyas and Cindy Grimm and Joseph R. Davidson},
url = {https://www.frontiersin.org/journals/robotics-and-ai/articles/10.3389/frobt.2025.1698591/full},
doi = {10.3389/frobt.2025.1698591},
issn = {2296-9144},
year = {2025},
date = {2025-12-01},
urldate = {2025-12-01},
journal = {Frontiers in Robotics and AI},
volume = {12},
publisher = {Frontiers},
abstract = {When a passively compliant hand grasps an object, slip events are often accompanied by flexion or extension of the finger or finger joints. This paper investigates whether a combination of orientation change and slip-induced vibration at the fingertip, as sensed by an inertial measurement unit (IMU), can be used as a slip indicator. Using a tendon-driven hand, which achieves passive compliance through underactuation, we performed 195 manipulation trials involving both slip and non-slip conditions. We then labeled this data automatically using motion-tracking data, and trained a convolutional neural network (CNN) to detect the slip events. Our results show that slip can be successfully detected from IMU data, even in the presence of other disturbances. This remains the case when deploying the trained network on data from a different gripper performing a new manipulation task on a previously unseen object.},
keywords = {AI, Labor},
pubstate = {published},
tppubtype = {article}
}
When a passively compliant hand grasps an object, slip events are often accompanied by flexion or extension of the finger or finger joints. This paper investigates whether a combination of orientation change and slip-induced vibration at the fingertip, as sensed by an inertial measurement unit (IMU), can be used as a slip indicator. Using a tendon-driven hand, which achieves passive compliance through underactuation, we performed 195 manipulation trials involving both slip and non-slip conditions. We then labeled this data automatically using motion-tracking data, and trained a convolutional neural network (CNN) to detect the slip events. Our results show that slip can be successfully detected from IMU data, even in the presence of other disturbances. This remains the case when deploying the trained network on data from a different gripper performing a new manipulation task on a previously unseen object.
4.
Alejandro Velasquez; Cindy Grimm; Joseph R. Davidson
Compact Robotic Gripper With Tandem Actuation for Selective Apple Harvesting Journal Article
In: IEEE Robotics and Automation Letters, vol. 10, no. 10, pp. 11030–11037, 2025, ISSN: 2377-3766.
Abstract | Links | BibTeX | Tags: AI, Labor
@article{velasquez_compact_2025,
title = {Compact Robotic Gripper With Tandem Actuation for Selective Apple Harvesting},
author = {Alejandro Velasquez and Cindy Grimm and Joseph R. Davidson},
url = {https://ieeexplore.ieee.org/document/11155195},
doi = {10.1109/LRA.2025.3608635},
issn = {2377-3766},
year = {2025},
date = {2025-10-01},
urldate = {2025-10-01},
journal = {IEEE Robotics and Automation Letters},
volume = {10},
number = {10},
pages = {11030\textendash11037},
abstract = {One of the primary reasons robotic apple harvesting is a challenging manipulation problem is the cluttered tree canopy. An effective harvesting gripper should i) be compact to minimize collisions with the canopy, ii) offer a compliant grasp to prevent bruising; and iii) hold the fruit securely to counteract forces during picking. Much of the prior work has used single-mode grippers (suction or fingers), which are often compliant but have low grasp strength (suction), or have a strong grasp but a large form factor (fingers). We present a compact robotic gripper that combines the benefits of both. It first uses an array of soft suction cups to gently attach to the fruit, then deploys three telescoping fingers that sweep away obstacles and pivot inward to secure the grasp. We analyze the finger design for its ability to sweep clutter and maintain a tight grasp, and we measure grasp strength across suction-only, fingers-only, and combined (tandem) actuation modes. Tandem mode consistently provides a grasp that can counter typically observed fruit detachment forces. Using an apple proxy, we test the gripper\'s performance in cluttered scenarios, achieving over 96% pick success with an ideal controller. Finally, we validate the gripper in a commercial apple orchard, achieving an 81% pick success rate.},
keywords = {AI, Labor},
pubstate = {published},
tppubtype = {article}
}
One of the primary reasons robotic apple harvesting is a challenging manipulation problem is the cluttered tree canopy. An effective harvesting gripper should i) be compact to minimize collisions with the canopy, ii) offer a compliant grasp to prevent bruising; and iii) hold the fruit securely to counteract forces during picking. Much of the prior work has used single-mode grippers (suction or fingers), which are often compliant but have low grasp strength (suction), or have a strong grasp but a large form factor (fingers). We present a compact robotic gripper that combines the benefits of both. It first uses an array of soft suction cups to gently attach to the fruit, then deploys three telescoping fingers that sweep away obstacles and pivot inward to secure the grasp. We analyze the finger design for its ability to sweep clutter and maintain a tight grasp, and we measure grasp strength across suction-only, fingers-only, and combined (tandem) actuation modes. Tandem mode consistently provides a grasp that can counter typically observed fruit detachment forces. Using an apple proxy, we test the gripper's performance in cluttered scenarios, achieving over 96% pick success with an ideal controller. Finally, we validate the gripper in a commercial apple orchard, achieving an 81% pick success rate.
5.
Tieqiao Wang; Abhinav Jain; Liqiang He; Cindy Grimm; Sinisa Todorovic
A Dataset for Semantic and Instance Segmentation of Modern Fruit Orchards Proceedings Article
In: 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), pp. 5381–5391, 2025, ISSN: 2160-7516, (ISSN: 2160-7516).
Abstract | Links | BibTeX | Tags: AI, Labor
@inproceedings{wang_dataset_2025,
title = {A Dataset for Semantic and Instance Segmentation of Modern Fruit Orchards},
author = {Tieqiao Wang and Abhinav Jain and Liqiang He and Cindy Grimm and Sinisa Todorovic},
url = {https://ieeexplore.ieee.org/document/11147695},
doi = {10.1109/CVPRW67362.2025.00535},
issn = {2160-7516},
year = {2025},
date = {2025-06-01},
urldate = {2025-06-01},
booktitle = {2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)},
pages = {5381\textendash5391},
abstract = {Automating orchard tasks, such as pruning tree branches, requires tree-structure understanding - a significant challenge for computer vision. This paper introduces the first large-scale dataset for semantic and instance segmentation of modern fruit orchards. It consists of videos showing Cherry and Apple trees in modern-orchard scenes, and includes both labeled synthetic and real data, along with synthetic tree meshes. To address prohibitive costs of annotating numerous tree branches, we study unsupervised domain adaptation from synthetic to real data. For this setting, we propose a new Semantically-Guided Depth Refinement (SGDR) that leverages zero-shot depth estimation and semantic-aware smoothing. SGDR outperforms strong baselines and state of the art. Furthermore, we also benchmark the dataset in the supervised setting, where the initial annotations from the first frame are automatically propagated throughout the video using the foundation Segment Anything Model (SAM). The resulting pseudo labels are then manually corrected to generate the ground truth. For the supervised setting, we introduce SAM-Mask2Former (SAM-M2F) aimed at instance segmentation. By providing this dataset and benchmarking for both settings, we aim to enable new research for precision agriculture.},
note = {ISSN: 2160-7516},
keywords = {AI, Labor},
pubstate = {published},
tppubtype = {inproceedings}
}
Automating orchard tasks, such as pruning tree branches, requires tree-structure understanding - a significant challenge for computer vision. This paper introduces the first large-scale dataset for semantic and instance segmentation of modern fruit orchards. It consists of videos showing Cherry and Apple trees in modern-orchard scenes, and includes both labeled synthetic and real data, along with synthetic tree meshes. To address prohibitive costs of annotating numerous tree branches, we study unsupervised domain adaptation from synthetic to real data. For this setting, we propose a new Semantically-Guided Depth Refinement (SGDR) that leverages zero-shot depth estimation and semantic-aware smoothing. SGDR outperforms strong baselines and state of the art. Furthermore, we also benchmark the dataset in the supervised setting, where the initial annotations from the first frame are automatically propagated throughout the video using the foundation Segment Anything Model (SAM). The resulting pseudo labels are then manually corrected to generate the ground truth. For the supervised setting, we introduce SAM-Mask2Former (SAM-M2F) aimed at instance segmentation. By providing this dataset and benchmarking for both settings, we aim to enable new research for precision agriculture.
6.
Abhinav Jain; Cindy Grimm; Stefan Lee
Learning to Prune Branches in Modern Tree-Fruit Orchards Proceedings Article
In: 2025 IEEE International Conference on Robotics and Automation (ICRA), pp. 15553–15559, 2025.
Abstract | Links | BibTeX | Tags: AI, Labor
@inproceedings{jain_learning_2025,
title = {Learning to Prune Branches in Modern Tree-Fruit Orchards},
author = {Abhinav Jain and Cindy Grimm and Stefan Lee},
url = {https://ieeexplore.ieee.org/document/11128361},
doi = {10.1109/ICRA55743.2025.11128361},
year = {2025},
date = {2025-05-01},
urldate = {2025-05-01},
booktitle = {2025 IEEE International Conference on Robotics and Automation (ICRA)},
pages = {15553\textendash15559},
abstract = {Dormant tree pruning is laborintensive but essential to maintaining modern highly-productive fruit orchards. In this work we present a closed-loop visuomotor controller for robotic pruning. The controller guides the cutter through a cluttered tree environment to reach a specified cut point and ensures the cutters are perpendicular to the branch. We train the controller using a novel orchard simulation that captures the geometric distribution of branches in a target apple orchard configuration. Unlike traditional methods requiring full 3D reconstruction, our controller uses just optical flow images from a wrist-mounted camera. We deploy our learned policy in simulation and the real-world for an example V-Trellis envy tree with zero-shot transfer, achieving a 30% success rate - approximately half the performance of an oracle planner.},
keywords = {AI, Labor},
pubstate = {published},
tppubtype = {inproceedings}
}
Dormant tree pruning is laborintensive but essential to maintaining modern highly-productive fruit orchards. In this work we present a closed-loop visuomotor controller for robotic pruning. The controller guides the cutter through a cluttered tree environment to reach a specified cut point and ensures the cutters are perpendicular to the branch. We train the controller using a novel orchard simulation that captures the geometric distribution of branches in a target apple orchard configuration. Unlike traditional methods requiring full 3D reconstruction, our controller uses just optical flow images from a wrist-mounted camera. We deploy our learned policy in simulation and the real-world for an example V-Trellis envy tree with zero-shot transfer, achieving a 30% success rate - approximately half the performance of an oracle planner.
7.
Martin Churuvija; Ranjan Sapkota; Dawood Ahmed; Manoj Karkee
A pose-versatile imaging system for comprehensive 3D modeling of planar-canopy fruit trees for automated orchard operations Journal Article
In: Computers and Electronics in Agriculture, vol. 230, pp. 109899, 2025, ISSN: 0168-1699.
Abstract | Links | BibTeX | Tags: AI, Labor
@article{churuvija_pose-versatile_2025,
title = {A pose-versatile imaging system for comprehensive 3D modeling of planar-canopy fruit trees for automated orchard operations},
author = {Martin Churuvija and Ranjan Sapkota and Dawood Ahmed and Manoj Karkee},
url = {https://www.sciencedirect.com/science/article/pii/S0168169925000055},
doi = {10.1016/j.compag.2025.109899},
issn = {0168-1699},
year = {2025},
date = {2025-03-01},
urldate = {2025-03-01},
journal = {Computers and Electronics in Agriculture},
volume = {230},
pages = {109899},
abstract = {High-density tree fruit production systems employ SNAP (Simple, Narrow, Accessible and Productive) canopy architectures, such as the UFO (Upright Fruiting Offshoots) system, that require intensive management practices. The growing adoption of these production systems in the USA, along with the decline of farm labor in the country, has sparked interest in automating manual orchard operations. Machine vision plays a key role in the development of robotic solutions because the success of these robots largely depends on the ability of the imaging systems (ISs) to quickly and accurately generate three-dimensional (3D) models of the surroundings. Tree models, for example, are essential to determine cutting points and guide cutting tools to the correct positions when pruning selectively. However, the ISs proposed in recent studies do not produce sufficiently comprehensive models, are expensive, and/or are impractical for commercial applications. In this study, a novel, time-efficient, and pose-versatile imaging system (Mobile IS) was developed and tested to overcome these issues. The Mobile IS utilized off-the-shelf cameras to capture an initial 3D point cloud model of a scene and then dynamically refined the model in real time by integrating additional point clouds from close range and different poses using simple photogrammetric techniques. To evaluate the performance of the Mobile IS, a wide range of UFO-trained tree offshoot diameters (OSDs), and side-branch lengths (SBLs) and spacings (SBSs)\textemdashparameters on which the pruning rules for UFOs are based\textemdashwere measured on the models reconstructed by the Mobile IS and a fixed-pose imaging system (Fixed IS) and compared to ground truth. Mobile IS models exhibited higher accuracy compared to the Fixed IS models, as evidenced by the root mean square (RMS) errors of the Mobile IS measurements (RMSEOSD of 4.9 mm, RMSESBL of 8.0 mm, and RMSESBS of 3.6 mm) and the Fixed IS measurements (RMSEOSD of 5.9 mm, RMSESBL of 18.1 mm, and RMSESBS of 3.8 mm). The versatility of pose enabled the Mobile IS to overcome occlusions and areas with low-confidence depth values. The results suggest that the Mobile IS holds promise as an IS for various robotic applications, including automated pruning, thinning and harvesting, across different tree fruit crops and canopy architectures.},
keywords = {AI, Labor},
pubstate = {published},
tppubtype = {article}
}
High-density tree fruit production systems employ SNAP (Simple, Narrow, Accessible and Productive) canopy architectures, such as the UFO (Upright Fruiting Offshoots) system, that require intensive management practices. The growing adoption of these production systems in the USA, along with the decline of farm labor in the country, has sparked interest in automating manual orchard operations. Machine vision plays a key role in the development of robotic solutions because the success of these robots largely depends on the ability of the imaging systems (ISs) to quickly and accurately generate three-dimensional (3D) models of the surroundings. Tree models, for example, are essential to determine cutting points and guide cutting tools to the correct positions when pruning selectively. However, the ISs proposed in recent studies do not produce sufficiently comprehensive models, are expensive, and/or are impractical for commercial applications. In this study, a novel, time-efficient, and pose-versatile imaging system (Mobile IS) was developed and tested to overcome these issues. The Mobile IS utilized off-the-shelf cameras to capture an initial 3D point cloud model of a scene and then dynamically refined the model in real time by integrating additional point clouds from close range and different poses using simple photogrammetric techniques. To evaluate the performance of the Mobile IS, a wide range of UFO-trained tree offshoot diameters (OSDs), and side-branch lengths (SBLs) and spacings (SBSs)—parameters on which the pruning rules for UFOs are based—were measured on the models reconstructed by the Mobile IS and a fixed-pose imaging system (Fixed IS) and compared to ground truth. Mobile IS models exhibited higher accuracy compared to the Fixed IS models, as evidenced by the root mean square (RMS) errors of the Mobile IS measurements (RMSEOSD of 4.9 mm, RMSESBL of 8.0 mm, and RMSESBS of 3.6 mm) and the Fixed IS measurements (RMSEOSD of 5.9 mm, RMSESBL of 18.1 mm, and RMSESBS of 3.8 mm). The versatility of pose enabled the Mobile IS to overcome occlusions and areas with low-confidence depth values. The results suggest that the Mobile IS holds promise as an IS for various robotic applications, including automated pruning, thinning and harvesting, across different tree fruit crops and canopy architectures.
8.
Dawood Ahmed; Basit Muhammad Imran; Martin Churuvija; Manoj Karkee
An Integrated Visual Servoing Framework for Precise Robotic Pruning Operations in Modern Commercial Orchard Journal Article
In: IFAC-PapersOnLine, vol. 59, no. 23, pp. 28–33, 2025, ISSN: 2405-8963.
Abstract | Links | BibTeX | Tags: AI, Labor
@article{ahmed_integrated_2025b,
title = {An Integrated Visual Servoing Framework for Precise Robotic Pruning Operations in Modern Commercial Orchard},
author = {Dawood Ahmed and Basit Muhammad Imran and Martin Churuvija and Manoj Karkee},
url = {https://www.sciencedirect.com/science/article/pii/S2405896325024565},
doi = {10.1016/j.ifacol.2025.11.758},
issn = {2405-8963},
year = {2025},
date = {2025-01-01},
urldate = {2025-01-01},
journal = {IFAC-PapersOnLine},
volume = {59},
number = {23},
pages = {28\textendash33},
series = {8th IFAC Conference on Sensing, Control and Automation Technologies for Agriculture AGRICONTROL 2025},
abstract = {This study presents a vision-guided robotic control system for automated fruit tree pruning applications. Traditional pruning practices are labor-intensive and limit agricultural efficiency and scalability, highlighting the need for advanced automation. A key challenge is the precise, robust positioning of the cutting tool in complex orchard environments, where dense branches and occlusions make target access difficult. To address this, an Intel RealSense D435 camera is mounted on the flange of a UR5e robotic arm and CoTracker3, a transformer-based point tracker, is utilized for visual servoing control that centers tracked points in the camera view. The system integrates proportional control with iterative inverse kinematics to achieve precise end-effector positioning. The system was validated in Gazebo simulation, achieving a 77.77% success rate within 5mm positional tolerance and 100% success rate within 10mm tolerance, with a mean end-effector error of 4.28 ± 1.36 mm. The vision controller demonstrated robust performance across diverse target positions within the pixel workspace. The results validate the effectiveness of integrating vision-based tracking with kinematic control for precision agricultural tasks. Future work will focus on real-world implementation and the integration of force sensing for actual cutting operations.},
keywords = {AI, Labor},
pubstate = {published},
tppubtype = {article}
}
This study presents a vision-guided robotic control system for automated fruit tree pruning applications. Traditional pruning practices are labor-intensive and limit agricultural efficiency and scalability, highlighting the need for advanced automation. A key challenge is the precise, robust positioning of the cutting tool in complex orchard environments, where dense branches and occlusions make target access difficult. To address this, an Intel RealSense D435 camera is mounted on the flange of a UR5e robotic arm and CoTracker3, a transformer-based point tracker, is utilized for visual servoing control that centers tracked points in the camera view. The system integrates proportional control with iterative inverse kinematics to achieve precise end-effector positioning. The system was validated in Gazebo simulation, achieving a 77.77% success rate within 5mm positional tolerance and 100% success rate within 10mm tolerance, with a mean end-effector error of 4.28 ± 1.36 mm. The vision controller demonstrated robust performance across diverse target positions within the pixel workspace. The results validate the effectiveness of integrating vision-based tracking with kinematic control for precision agricultural tasks. Future work will focus on real-world implementation and the integration of force sensing for actual cutting operations.
9.
Ranjan Sapkota; Manoj Karkee
Improved YOLOv12 with LLM-Generated Synthetic Data for Enhanced Apple Detection and Benchmarking Against YOLOv11 and YOLOv10 Journal Article
In: IFAC-PapersOnLine, vol. 59, no. 23, pp. 239–244, 2025, ISSN: 2405-8963.
Abstract | Links | BibTeX | Tags: AI, Labor
@article{sapkota_improved_2025b,
title = {Improved YOLOv12 with LLM-Generated Synthetic Data for Enhanced Apple Detection and Benchmarking Against YOLOv11 and YOLOv10},
author = {Ranjan Sapkota and Manoj Karkee},
url = {https://www.sciencedirect.com/science/article/pii/S2405896325024929},
doi = {10.1016/j.ifacol.2025.11.793},
issn = {2405-8963},
year = {2025},
date = {2025-01-01},
urldate = {2025-01-01},
journal = {IFAC-PapersOnLine},
volume = {59},
number = {23},
pages = {239\textendash244},
series = {8th IFAC Conference on Sensing, Control and Automation Technologies for Agriculture AGRICONTROL 2025},
abstract = {Accurate fruit detection in orchards remains a major bottleneck in agricultural automation due to the high cost and labor demands of data collection and annotation. This study evaluated the performance of the YOLOv12 object detection model, and compared against the performances YOLOv11 and YOLOv10 for apple detection in commercial orchards based on the model training completed entirely on synthetic images generated by Large Language Models (LLMs). The YOLOv12n configuration achieved the highest precision at 0.916, the highest recall at 0.969, and the highest mean Average Precision (mAP@50) at 0.978. In comparison, the YOLOv11 series was led by YOLO11x, which achieved the highest precision at 0.857, recall at 0.85, and mAP@50 at 0.91. For the YOLOv10 series, YOLOv10b and YOLOv10l both achieved the highest precision at 0.85, with YOLOv10n achieving the highest recall at 0.8 and mAP@50 at 0.89. These findings demonstrated that YOLOv12, when trained on realistic LLM-generated datasets surpassed its predecessors in key performance metrics. The technique also offered a cost-effective solution by reducing the need for extensive manual data collection in the agricultural field. In addition, this study compared the computational efficiency of all versions of YOLOv12, v11 and v10, where YOLOv11n reported the lowest inference time at 4.7 ms, compared to YOLOv12n’s 5.6 ms and YOLOv10n’s 5.9 ms. Although YOLOv12 is new and more accurate than YOLOv11, and YOLOv10, YOLO11n still stays the fastest YOLO model among YOLOv10, YOLOv11 and YOLOv12 series of models.},
keywords = {AI, Labor},
pubstate = {published},
tppubtype = {article}
}
Accurate fruit detection in orchards remains a major bottleneck in agricultural automation due to the high cost and labor demands of data collection and annotation. This study evaluated the performance of the YOLOv12 object detection model, and compared against the performances YOLOv11 and YOLOv10 for apple detection in commercial orchards based on the model training completed entirely on synthetic images generated by Large Language Models (LLMs). The YOLOv12n configuration achieved the highest precision at 0.916, the highest recall at 0.969, and the highest mean Average Precision (mAP@50) at 0.978. In comparison, the YOLOv11 series was led by YOLO11x, which achieved the highest precision at 0.857, recall at 0.85, and mAP@50 at 0.91. For the YOLOv10 series, YOLOv10b and YOLOv10l both achieved the highest precision at 0.85, with YOLOv10n achieving the highest recall at 0.8 and mAP@50 at 0.89. These findings demonstrated that YOLOv12, when trained on realistic LLM-generated datasets surpassed its predecessors in key performance metrics. The technique also offered a cost-effective solution by reducing the need for extensive manual data collection in the agricultural field. In addition, this study compared the computational efficiency of all versions of YOLOv12, v11 and v10, where YOLOv11n reported the lowest inference time at 4.7 ms, compared to YOLOv12n’s 5.6 ms and YOLOv10n’s 5.9 ms. Although YOLOv12 is new and more accurate than YOLOv11, and YOLOv10, YOLO11n still stays the fastest YOLO model among YOLOv10, YOLOv11 and YOLOv12 series of models.
10.
Miranda Cravetz; Joseph R. Davidson
Abscission Joint Localization During Robotic Fruit Harvesting using Force Sensing Journal Article
In: IFAC-PapersOnLine, vol. 59, no. 23, pp. 425–430, 2025, ISSN: 2405-8963.
Abstract | Links | BibTeX | Tags: AI, Labor
@article{cravetz_abscission_2025,
title = {Abscission Joint Localization During Robotic Fruit Harvesting using Force Sensing},
author = {Miranda Cravetz and Joseph R. Davidson},
url = {https://www.sciencedirect.com/science/article/pii/S2405896325025315},
doi = {10.1016/j.ifacol.2025.11.825},
issn = {2405-8963},
year = {2025},
date = {2025-01-01},
urldate = {2025-01-01},
journal = {IFAC-PapersOnLine},
volume = {59},
number = {23},
pages = {425\textendash430},
series = {8th IFAC Conference on Sensing, Control and Automation Technologies for Agriculture AGRICONTROL 2025},
abstract = {Fruit separation techniques are often sensitive to the location of the fruit’s peduncle, which is the fruit-bearing stem of the plant. Most work on peduncle or abscission joint localization has focused on computer vision, but such features can be difficult to visually access due to the cluttered nature of agricultural environments. Our work proposes an alternative method which relies on mechanical \textemdash rather than visual \textemdash information to localize the fruit’s abscission joint. Using laboratory data gathered from our physical orchard proxy, we demonstrate that the technique is able to localize the abscission joint within a median distance of 3.8 cm. We also evaluate our technique in commercial orchards using two different gripper designs.},
keywords = {AI, Labor},
pubstate = {published},
tppubtype = {article}
}
Fruit separation techniques are often sensitive to the location of the fruit’s peduncle, which is the fruit-bearing stem of the plant. Most work on peduncle or abscission joint localization has focused on computer vision, but such features can be difficult to visually access due to the cluttered nature of agricultural environments. Our work proposes an alternative method which relies on mechanical — rather than visual — information to localize the fruit’s abscission joint. Using laboratory data gathered from our physical orchard proxy, we demonstrate that the technique is able to localize the abscission joint within a median distance of 3.8 cm. We also evaluate our technique in commercial orchards using two different gripper designs.
