In the evolution from Agriculture 2.0 to Agriculture 5.0, the agricultural sector has witnessed transformative changes. Agriculture 5.0 represents a revolutionary phase, integrating advanced technologies to enhance efficiency and sustainability adapted to individual field and livestock needs. The integration of robotics, extended reality, and 6G technologies marks a significant leap, enabling real-time monitoring and automation in farming practices. Artificial intelligence and big data are pivotal in Agriculture 5.0, offering insights for decision making and predictive analytics. Natural language processing plays a crucial role in facilitating efficient communication and data processing. The transition to Agriculture 5.0 should also consider societal challenges, in terms of technology lock-ins and the need for behavior shifts among stakeholders for faster adoption and customization of the solutions. This article presents a comprehensive description of Agriculture 5.0, underscoring the potential and hurdles in reshaping the future of farming.

Robotics has been increasingly relevant over the years. The ever-increasing demand for productivity, the reduction of tedious labor, and safety for the operator and the environment have brought robotics to the forefront of technological innovation. The same principle applies to agricultural robots, where such solutions can aid in making farming easier for the farmers, safer, and with greater margins for profit, while at the same time offering higher quality products with minimal environmental impact. This paper focuses on reviewing the existing state of the art for vision-based perception in agricultural robots across a variety of field operations; specifically: weed detection, crop scouting, phenotyping, disease detection, vision-based navigation, harvesting, and spraying. The review revealed a large interest in the uptake of vision-based solutions in agricultural robotics, with RGB cameras being the most popular sensor of choice. It also outlined that AI can achieve promising results and that there is not a single algorithm that outperforms all others; instead, different artificial intelligence techniques offer their unique advantages to address specific agronomic problems.

Wine grapes need frequent monitoring to achieve high yields and quality. Non-destructive methods, such as proximal and remote sensing, are commonly used to estimate crop yield and quality characteristics, and spectral vegetation indices (VIs) are often used to present site-specific information. Analysis of laboratory samples is the most popular method for determining the quality characteristics of grapes, although it is time-consuming and expensive. In recent years, several machine learning-based methods have been developed to predict crop quality. Although these techniques require the extensive involvement of experts, automated machine learning (AutoML) offers the possibility to improve this task, saving time and resources. In this paper, we propose an innovative approach for robust prediction of grape quality attributes by combining open-source AutoML techniques and Normalized Difference Vegetation Index (NDVI) data for vineyards obtained from four different platforms-two proximal vehicle-mounted canopy reflectance sensors, orthomosaics from UAV images and Sentinel-2 remote sensing imagery-during the 2019 and 2020 growing seasons. We investigated AutoML, extending our earlier work on manually fine-tuned machine learning methods. Results of the two approaches using Ordinary Least Square (OLS), Theil-Sen and Huber regression models and tree-based methods were compared. Support Vector Machines (SVMs) and Automatic Relevance Determination (ARD) were included in the analysis and different combinations of sensors and data collected over two growing seasons were investigated. Results showed promising performance of Unmanned Aerial Vehicle (UAV) and Spectrosense+ GPS data in predicting grape sugars, especially in mid to late season with full canopy growth. Regression models with both manually fine-tuned ML (R² = 0.61) and AutoML (R² = 0.65) provided similar results, with the latter slightly improved for both 2019 and 2020. When combining multiple sensors and growth stages per year, the coefficient of determination R² improved even more averaging 0.66 for the best-fitting regressions. Also, when considering combinations of sensors and growth stages across both cropping seasons, UAV and Spectrosense+ GPS, as well as Véraison and Flowering, each had the highest average R² values. These performances are consistent with previous work on machine learning algorithms that were manually fine-tuned. These results suggest that AutoML has greater long-term performance potential. To increase the efficiency of crop quality prediction, a balance must be struck between manual expert work and AutoML.

Broccoli is an example of a high-value crop that requires delicate handling throughout the growing season and during its post-harvesting treatment. As broccoli heads can be easily damaged, they are still harvested by hand. Moreover, human scouting is required to initially identify the field segments where several broccoli plants have reached the desired maturity level, such that they can be harvested while they are in the optimal condition. The aim of this study was to automate this process using state-of-the-art Object Detection architectures trained on georeferenced orthomosaic-derived RGB images captured from low-altitude UAV flights, and to assess their capacity to effectively detect and classify broccoli heads based on their maturity level. The results revealed that the object detection approach for automated maturity classification achieved comparable results to physical scouting overall, especially for the two best-performing architectures, namely Faster R-CNN and CenterNet. Their respective performances were consistently over 80% mAP@50 and 70% mAP@75 when using three levels of maturity, and even higher when simplifying the use case into a two-class problem, exceeding 91% and 83%, respectively. At the same time, geometrical transformations for data augmentations reported improvements, while colour distortions were counterproductive. The best-performing architecture and the trained model could be tested as a prototype in real-time UAV detections in order to assist in on-field broccoli maturity detection.

In this paper, we propose an innovative approach for robust prediction of processing tomato yield using open-source AutoML techniques and statistical analysis. Sentinel-2 satellite imagery was deployed to obtain values of five (5) selected vegetation indices (VIs) during the growing season of 2021 (April to September) at 5-day intervals. Actual recorded yields were collected across 108 fields, corresponding to a total area of 410.10 ha of processing tomato in central Greece, to assess the performance of Vis at different temporal scales. In addition, VIs were connected with the crop phenology to establish the annual dynamics of the crop. The highest Pearson coefficient (r) values occurred during a period of 80 to 90 days, indicating the strong relationship between the VIs and the yield. Specifically, RVI presented the highest correlation values of the growing season at 80 (r = 0.72) and 90 days (r = 0.75), while NDVI performed better at 85 days (r = 0.72). This output was confirmed by the AutoML technique, which also indicated the highest performance of the VIs during the same period, with the values of the adjusted R² ranging from 0.60 to 0.72. The most precise results were obtained with the combination of ARD regression and SVR, which was the most successful combination for building an ensemble (adj. R² = 0.67 ± 0.02).

In the past years, several machine-learning-based techniques have arisen for providing effective crop protection. For instance, deep neural networks have been used to identify different types of weeds under different real-world conditions. However, these techniques usually require extensive involvement of experts working iteratively in the development of the most suitable machine learning system. To support this task and save resources, a new technique called Automated Machine Learning has started being studied. In this work, a complete open-source Automated Machine Learning system was evaluated with two different datasets, (i) The Early Crop Weeds dataset and (ii) the Plant Seedlings dataset, covering the weeds identification problem. Different configurations, such as the use of plant segmentation, the use of classifier ensembles instead of Softmax and training with noisy data, have been compared. The results showed promising performances of 93.8% and 90.74% F₁ score depending on the dataset used. These performances were aligned with other related works in AutoML, but they are far from machine-learning-based systems manually fine-tuned by human experts. From these results, it can be concluded that finding a balance between manual expert work and Automated Machine Learning will be an interesting path to work in order to increase the efficiency in plant protection.