Article
Agricultural Engineering
Rekha Raja, David C. Slaughter, Steven A. Fennimore, Mark C. Siemens
Summary: The use of automated technology in agriculture has made it possible to automate tasks in a semi-structured, natural farming environment. This article proposes a fast-intelligent weed control system that utilizes crop signalling, machine vision, and a precision micro-jet sprayer to target and apply herbicide to in-row weeds with high accuracy and speed. The system successfully detected and sprayed weeds located between lettuce plants, achieving a 98% accuracy rate.
BIOSYSTEMS ENGINEERING
(2023)
Article
Agriculture, Multidisciplinary
Thijs Ruigrok, Eldert J. van Henten, Gert Kootstra
Summary: This paper investigates the effect of dataset distribution on the generalization error of plant-detection models and proposes incremental training as a solution. The YOLOv3 object detector is used as the plant-detection model, and a diverse dataset with variations in geographic areas, soil types, cultivation conditions, weeds, vegetation, camera quality, and illumination is used. The results show that increasing the number of sub-datasets and training images can improve generalization, but incremental training is necessary to adapt the model to specific scenarios outside the training distribution.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2023)
Article
Soil Science
C. MacLaren, J. Labuschagne, P. A. Swanepoel
Summary: Reduced tillage practices are generally considered more sustainable than conventional tillage practices, but controlling weeds remains a challenge for many producers. Crop rotation is often recommended for weed management in reduced tillage systems, but uncertainties exist about how different tillage practices and crop rotations interact. Our study in South Africa's winter rainfall region found that different tillage practices did not significantly affect weed density in wheat monoculture. Both crop rotations generally had lower weed densities and reduced dominance of grass weeds compared to monoculture, but zero tillage with crop rotation showed similar weed seed bank densities to monoculture, suggesting an antagonistic relationship in this system. Producers seeking to reduce tillage in the region should opt for minimum tillage over zero tillage and avoid wheat monoculture, while weed researchers and agronomists should be cautious of potential antagonistic interactions between weed management practices in different systems.
SOIL & TILLAGE RESEARCH
(2021)
Review
Agronomy
Marwan Albahar
Summary: The objective of this study was to provide a comprehensive overview of the recent advancements in the use of deep learning (DL) in the agricultural sector. The author conducted a review of studies published between 2016 and 2022 to highlight the various applications of DL in agriculture. DL shows great promise in transforming the agriculture industry, but challenges such as dataset compilation, computational power cost, and shortage of experts need to be addressed.
Review
Agriculture, Multidisciplinary
Nitin Rai, Yu Zhang, Billy G. Ram, Leon Schumacher, Ravi K. Yellavajjala, Sreekala Bajwa, Xin Sun
Summary: Deep Learning (DL) is transforming weed detection by integrating ground and aerial-based technologies to identify weeds in still images and real-time settings. A review of 60 technical papers on DL-based weed detection found that transfer learning is a widely adopted technique, custom designed neural networks are less focused on, and no specific model has achieved high accuracy on multiple field images. The review also highlighted the lack of research in optimizing models for resource-constrained devices and exploring ways to design efficient models with less training hours and parameters.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2023)
Article
Ecology
Sanjay Kumar Gupta, Shivam Kumar Yadav, Sanjay Kumar Soni, Udai Shanker, Pradeep Kumar Singh
Summary: This study proposes an automated approach for multiclass weed identification using semantic segmentation to improve weed control techniques, reduce pesticide usage, and enhance crop yields. A novel multiclass weed dataset was created and four advanced deep learning models were evaluated, with the U-Net-based Inception-ReseNetV2 achieving the highest F1-score of 96.78%. These findings demonstrate the efficacy of the proposed approach in accurately identifying and categorizing weeds in agricultural fields.
ECOLOGICAL INFORMATICS
(2023)
Review
Agriculture, Multidisciplinary
Marco Esposito, Mariano Crimaldi, Valerio Cirillo, Fabrizio Sarghini, Albino Maggio
Summary: Weeds are a significant abiotic factor impacting agriculture globally, causing important yield loss. Integrated Weed Management with the use of drones enables efficient and environmentally beneficial Site-Specific Weed Management. The identification of weed patches through drone image acquisition and machine learning techniques can lead to the training of specific algorithms for weed removal by Autonomous Weeding Robots.
CHEMICAL AND BIOLOGICAL TECHNOLOGIES IN AGRICULTURE
(2021)
Article
Computer Science, Information Systems
Muhammad Tufail, Javaid Iqbal, Mohsin Islam Tiwana, Muhammad Shahab Alam, Zubair Ahmad Khan, Muhammad Tahir Khan
Summary: This study presents a machine-learning based crop/weed detection system for a tractor-mounted boom sprayer, achieving accurate classification while ensuring real-time inference. The SVM classifier performs well in terms of accuracy (96%) and real-time inference (6 FPS) on an embedded device (Raspberry Pi 4), outperforming a customized deep learning-based classifier in terms of speed.
Article
Environmental Sciences
Eduardo Assuncao, Pedro D. Gaspar, Ricardo Mesquita, Maria P. Simoes, Khadijeh Alibabaei, Andre Veiros, Hugo Proenca
Summary: This paper investigates the impact of model optimization on segmentation performance in edge devices. The experimental results show significant acceleration in inference time but a decrease in segmentation performance after optimizing the model. The authors also describe an application of semantic segmentation of weeds embedded in an edge device integrated with a robotic orchard.
Article
Plant Sciences
Talha Ilyas, Jonghoon Lee, Okjae Won, Yongchae Jeong, Hyongsuk Kim
Summary: Recent developments in deep learning-based automatic weeding systems have shown promise for unmanned weed eradication. However, accurately distinguishing between crops and weeds in varying field conditions remains a challenge for these systems. In this study, an approach based on unsupervised domain adaptation was proposed to improve crop-weed recognition in new, unseen fields. The approach addresses the issue of performance deterioration caused by insignificant changes in low-level statistics and a gap between training and test data distributions. The proposed method demonstrated consistent improvements in performance on four different unseen fields.
FRONTIERS IN PLANT SCIENCE
(2023)
Article
Automation & Control Systems
Mulham Fawakherji, Ciro Potena, Alberto Pretto, Domenico D. Bloisi, Daniele Nardi
Summary: This study proposes an alternative approach to data augmentation using GANs to improve crop/weed segmentation in precision farming. By generating semi-artificial samples and multi-spectral synthetic images with a conditional GAN, the model is shown to produce realistic plant images and enhance the performance of semantic segmentation convolutional networks.
ROBOTICS AND AUTONOMOUS SYSTEMS
(2021)
Article
Agronomy
Lorenzo Gagliardi, Mino Sportelli, Marco Fontanelli, Massimo Sbrana, Sofia Matilde Luglio, Michele Raffaelli, Andrea Peruzzi
Summary: Conservation agriculture practices, such as reduced tillage and the incorporation of cover crops, are important for improving the sustainability of organic farming systems. This two-year field trial evaluated different organic itineraries with varying soil management and weed control strategies for tomato cultivation. The use of biodegradable mulch and mechanical weeding showed promising results in terms of weed biomass and tomato yield. However, these strategies also resulted in higher costs, highlighting the need for further research on the long-term effects and economic advantages of shallower soil tillage and cover crop management.
Article
Plant Sciences
Judit Barroso, Nicholas G. Genna
Summary: Russian thistle is a persistent issue for farmers in the Pacific Northwest, requiring integrated management strategies for control. Research in Oregon showed that spring barley had lower emergence and higher mortality of Russian thistle compared to spring wheat. Row spacing and seeding rate had little to no effect on Russian thistle emergence or mortality, with crop yield being the main factor influencing weed biomass and seed production. Increasing seeding rates or planting in narrow rows may improve crop yield in low rainfall years, but no significant effect was observed in higher rainfall years.
Article
Agriculture, Multidisciplinary
Halil Mertkan Sahin, Tajul Miftahushudur, Bruce Grieve, Hujun Yin
Summary: Weeds pose a major challenge to agriculture, competing with crops for resources and causing significant yield losses. Early detection is crucial for taking appropriate action, and this study introduces a deep learning model for weed detection. The proposed method offers a viable approach for early-stage weed detection.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2023)
Article
Plant Sciences
Chunshi Nong, Xijian Fan, Junling Wang
Summary: Weed control is crucial for crop yield and food production. This paper proposes a method called SemiWeedNet, which uses semi-supervised learning to accurately identify weeds of varying sizes in complex environments. Experimental results demonstrate that SemiWeedNet outperforms existing methods and has great potential in improving image segmentation.
FRONTIERS IN PLANT SCIENCE
(2022)
Article
Robotics
Petra Bosilj, Erchan Aptoula, Tom Duckett, Grzegorz Cielniak
JOURNAL OF FIELD ROBOTICS
(2020)
Article
Computer Science, Artificial Intelligence
Zhi Yan, Tom Duckett, Nicola Bellotto
Article
Robotics
Jaime Pulido Fentanes, Amir Badiee, Tom Duckett, Jonathan Evans, Simon Pearson, Grzegorz Cielniak
JOURNAL OF FIELD ROBOTICS
(2020)
Article
Chemistry, Analytical
Raymond Kirk, Grzegorz Cielniak, Michael Mangan
Article
Agricultural Engineering
Petra Bosilj, Iain Gould, Tom Duckett, Grzegorz Cielniak
BIOSYSTEMS ENGINEERING
(2020)
Article
Robotics
Adrian Salazar Gomez, Erchan Aptoula, Simon Parsons, Petra Bosilj
Summary: This study compared three methods for counting fruit or grains in images, showing that as object density increases, counting by regression becomes more accurate than counting by detection. The error in count predicted by detection-based methods can be up to 5 times higher than regression-based methods when there are more than a hundred objects per image.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Environmental Sciences
Amir Badiee, John R. Wallbank, Jaime Pulido Fentanes, Emily Trill, Peter Scarlet, Yongchao Zhu, Grzegorz Cielniak, Hollie Cooper, James R. Blake, Jonathan G. Evans, Marek Zreda, Markus Koehli, Simon Pearson
Summary: The study explores the use of high-density polyethylene moderator to limit the footprint of a soil moisture sensor, improving localization of moisture variation in the field. Results show that additional moderator can significantly reduce the sensor's footprint, double the percentage of detected neutrons within 5 meters, and sense moisture changes over smaller length scales.
WATER RESOURCES RESEARCH
(2021)
Article
Robotics
Sergi Molina, Grzegorz Cielniak, Tom Duckett
Summary: Understanding human activity patterns is crucial for efficient robot navigation in human environments. This article introduces a new mobile robot exploration methodology that maximizes knowledge of human activity patterns by deciding where and when to collect observations, with results showing improved prediction accuracy in certain scenarios. The exploration ratio is identified as a key factor in model prediction accuracy.
IEEE TRANSACTIONS ON ROBOTICS
(2022)
Article
Robotics
Jose Carlos Mayoral Banos, Pal Johan From, Grzegorz Cielniak
Summary: Safe navigation is crucial for autonomous applications, especially those involving mobile tasks, to prevent dangerous situations and harm to humans. However, integrating a risk management process is not yet mandatory in robotics development. Ensuring safety in real-world applications, such as agricultural environments using mobile devices with industrial cutters, is critical. This paper proposes the explicit integration of a risk management process into the software design of an autonomous grass mower to enhance safety. The approach is tested and validated in simulated scenarios, assessing the effectiveness of custom safety features in terms of collision prevention, execution time, and required human intervention.
Article
Robotics
Riccardo Polvara, Sergi Molina, Ibrahim Hroob, Alexios Papadimitriou, Konstantinos Tsiolis, Dimitrios Giakoumis, Spiridon Likothanassis, Dimitrios Tzovaras, Grzegorz Cielniak, Marc Hanheide
Summary: Achieving a robust long-term deployment with mobile robots in agriculture is challenging due to the continuously changing environment. In this study, an ongoing effort in the long-term deployment of an autonomous mobile robot in a vineyard is reported, with the objective of acquiring a data set for testing mapping and localization algorithms. The data set covers a total of 7 months and captures the canopy growth from March to September. An initial study on long-term localization using different sessions belonging to different months and plant stages is also presented.
JOURNAL OF FIELD ROBOTICS
(2023)
Article
Robotics
Rajitha de Silva, Grzegorz Cielniak, Gang Wang, Junfeng Gao
Summary: This paper presents a robust crop row detection algorithm using inexpensive cameras to withstand field variations in agricultural environments. A data set of sugar beet images representing 11 field variations was created for testing. The proposed algorithm segments the crop rows using deep learning and extracts the central crop row with a novel selection algorithm. The algorithm demonstrates robust vision-based crop row detection in challenging field conditions, outperforming the baseline.
JOURNAL OF FIELD ROBOTICS
(2023)
Review
Biodiversity Conservation
Katherine Margaret Frances James, Daniel James Sargent, Adam Whitehouse, Grzegorz Cielniak
Summary: This review assesses the current status and future potential of automated phenotyping in strawberry crops, highlighting key advances and the challenges that need to be addressed. While automated assessment of external morphological traits in strawberries is important for breeding, there are still limitations when applying high-throughput phenotyping in real-world conditions.
PLANTS PEOPLE PLANET
(2022)
Article
Computer Science, Interdisciplinary Applications
Francesco Pistolesi, Michele Baldassini, Beatrice Lazzerini
Summary: More than one in four workers worldwide suffer from back pain, resulting in the loss of 264 million work days annually. In the U.S., it costs $50 billion in healthcare expenses each year, rising up to $100 billion when accounting for decreased productivity and lost wages. The impending Industry 5.0 revolution emphasizes worker well-being and their rights, such as privacy, autonomy, and human dignity. This paper proposes a privacy-preserving artificial intelligence system that monitors the posture of assembly line workers. The system accurately assesses upper-body and lower-body postures while respecting privacy, enabling the detection of harmful posture habits and reducing the likelihood of musculoskeletal disorders.
COMPUTERS IN INDUSTRY
(2024)
Article
Computer Science, Interdisciplinary Applications
Xavier Boucher, Camilo Murillo Coba, Damien Lamy
Summary: This paper explores the new business strategies of digital servitization and smart PSS delivery, and develops conceptual prototypes of smart PSS value offers for early stages of the design process. It presents the development and experimentation of a modelling language and toolkit, and applies it to the design of a smart PSS in the field of heating appliances.
COMPUTERS IN INDUSTRY
(2024)
Article
Computer Science, Interdisciplinary Applications
Dieudonne Tchuente, Jerry Lonlac, Bernard Kamsu-Foguem
Summary: Artificial Intelligence (AI) is becoming increasingly important in various sectors of society. However, the black box nature of most AI techniques such as Machine Learning (ML) hinders their practical application. This has led to the emergence of Explainable artificial intelligence (XAI), which aims to provide AI-based decision-making processes and outcomes that are easily understood, interpreted, and justified by humans. While there has been a significant amount of research on XAI, there is currently a lack of studies on its practical applications. To address this research gap, this article proposes a comprehensive review of the business applications of XAI and a six-step framework to improve its implementation and adoption by practitioners.
COMPUTERS IN INDUSTRY
(2024)
Article
Computer Science, Interdisciplinary Applications
Francois-Alexandre Tremblay, Audrey Durand, Michael Morin, Philippe Marier, Jonathan Gaudreault
Summary: Continuous high-frequency wood drying, integrated with a traditional wood finishing line, improves the value of lumber by correcting moisture content piece by piece. Using reinforcement learning for continuous drying operation policies outperforms current industry methods and remains robust to sudden disturbances.
COMPUTERS IN INDUSTRY
(2024)
Article
Computer Science, Interdisciplinary Applications
Luyao Xia, Jianfeng Lu, Yuqian Lu, Wentao Gao, Yuhang Fan, Yuhao Xu, Hao Zhang
Summary: Efficient assembly sequence planning is crucial for enhancing production efficiency, ensuring product quality, and meeting market demands. This study proposes a dynamic graph learning algorithm called assembly-oriented graph attention sequence (A-GASeq), which optimizes the assembly graph structure to guide the search for optimal assembly sequences. The algorithm demonstrates superiority and broad utility in real-world scenarios.
COMPUTERS IN INDUSTRY
(2024)
Article
Computer Science, Interdisciplinary Applications
Mutahar Safdar, Padma Polash Paul, Guy Lamouche, Gentry Wood, Max Zimmermann, Florian Hannesen, Christophe Bescond, Priti Wanjara, Yaoyao Fiona Zhao
Summary: Metal-based additive manufacturing can achieve fully dense metallic components, and the application of machine learning in this field has been growing rapidly. However, there is a lack of framework to manage these machine learning models and guidance on the fundamental requirements for a cross-disciplinary platform to support process-based machine learning models in industrial metal AM.
COMPUTERS IN INDUSTRY
(2024)