Journal of Ambient Intelligence and Smart Environments

This paper focuses on the prospects and challenges of emerging agricultural technologies and provides a comprehensive overview of the current state of agricultural technology. This review paper examines the potential benefits and risks of robotics, artificial intelligence (AI), and 5G technology applications in agriculture. It provides a comprehensive overview of the current state of Agricultural Technology trends, including the most promising applications. This paper highlights the use of robots, drones and AI algorithms in precision agriculture, crop monitoring, autonomous agriculture, live-stock monitoring, and farm-to-table logistics. The challenge, however, is a lack of reliable infrastructure, effective data management, and a clear regulatory framework. The study comes to the conclusion that although AgriTech has the potential to increase agricultural sustainability and production, its use will present some difficulties. AgriTech innovations can only be fully realized by inculcating more research and development.

Forests are crucial for preserving biodiversity and regulating the global climate. However, they are increasingly at risk from destructive wildfires that threaten the environment and human communities. Accurate prediction models are essential to minimize the impact of forest fires. This study presents a new hybrid model that combines the Apriori association rule mining algorithm with the binary golden ratio optimization method (BGROM) to improve the accuracy of forest fire prediction. The BGROM, based on the golden ratio observed in plant and animal growth and formulated by the renowned mathematician Fibonacci. It is used to select the candidate features, which are then used by the Apriori algorithm to generate classification rules to predict the risk of wildfires. Integrating the Apriori algorithm with BGROM improves the accuracy of forest fire prediction and enhances our understanding of the complex interactions and patterns that influence wildfire behavior. This innovative approach holds great promise for advancing the development of effective forest fire prevention and management strategies. Experimental results show that the proposed model outperforms existing prediction methods, offering a more reliable tool for early forest fire detection and risk management.

Maintaining thermal comfort and regulating humidity in footwear is critical, particularly in environments where fluctuations in temperature and moisture impact user experience. Correct estimation of these parameters is crucial in selecting suitable material, thereby improving the efficiency and comfort of the footwear. This study proposes a comparison of the deep learning models (recurrent neural networks, long short-term memory (LSTM), bidirectional LSTM, DeepAR, and hybrid model (Cay, Vassiliadis et al.)) for the forecasting of temperature and humidity inside various types of footwear based on the multi-sensor data. A foot model that mimicked the actual temperature and perspiration process of the human foot was used, and sensor readings were taken from various points of the footwear. Performance of deep learning models was evaluated using four key metrics: mean absolute error, root mean squared error, explained variance score, and mean absolute percentage error. The findings indicate that the hybrid model outperforms the other models and achieves the highest predictive accuracy across all tested footwear conditions. This research contributes to the development of artificial intelligence-based predictive modeling for footwear climate control, offering a robust approach for determining thermophysiological comfort in wearable technologies and smart-manufacturing applications.

Graphical Abstract

This is a visual representation of the abstract.

By addressing the operational and maintenance constraints of physical ground-based weather stations, this study proposes a deep learning (DL) framework for estimating reference evapotranspiration (ET₀) by combining open-access climate services and remote sensing (RS) data. The proposed approach is benchmarked against traditional machine learning (ML) models, while multiple deep neural network (DNN) architectures are also evaluated, including multilayer perceptron (MLP), convolutional neural networks (CNNs), and recurrent neural networks (RNNs). Experiments conducted on three agricultural plots in southeastern Spain, representing contrasting meteorological conditions, demonstrate that RNNs achieve the best performance, with a coefficient of determination of R² = 0.92. Furthermore, model interpretability was addressed using SHapley Additive exPlanations analysis, which confirmed the biophysical consistency of the predictions and identified land surface temperature as the primary driver of the model's estimations. A key contribution is the demonstration that infrastructure-free models trained solely on open-access satellite and climate data can match or even surpass conventional meteorology-based methods, providing a scalable solution for ET₀ prediction. Based on a validation performed upon three agricultural plots in southeastern Spain, which represent contrasting semi-arid meteorological conditions, the framework showcases its potential applicability for global scalability by utilizing location-agnostic, open-access data. Moreover, the integration of crop coefficients enables accurate forecasting of daily irrigation demand. Overall, the proposed methodology illustrates the feasibility of artificial intelligence-driven irrigation management across diverse climates and highlights its potential to advance sustainable water use in agriculture.

Serverless computing offers automatic resource management and pay-per-use execution, but autoscaling remains difficult due to cold-start latency, inter-function dependencies, and highly dynamic workloads. Many existing approaches scale functions independently or rely on a single predictor, which can reduce robustness and cost efficiency. We present a dependency-aware autoscaling framework that unifies bottleneck identification, short-horizon demand forecasting, and cost-aware control in an end-to-end pipeline. We model applications as directed dependency graphs and prioritize high-impact functions using degree centrality. For these bottlenecks, near-term demand is predicted using lightweight supervised models, whose outputs are fused via a performance-weighted probabilistic ensemble inspired by Bayesian model averaging to improve stability under workload variability. The controller also accounts for cold starts and filters candidate actions through a cost-comparison mechanism to balance latency and operational efficiency. Experiments on real workload traces show improved prediction accuracy and more stable scaling decisions than representative baselines; supervised forecasting also consistently outperforms unsupervised clustering for generating autoscaling actions. The primary contribution is a practical system-level design that integrates dependency analysis, ensemble-based prediction, and cost-aware decision-making for robust serverless autoscaling.

The present study aimed to test the accuracy of applying machine learning to a novel contactless video-based approach in detecting task-related concentration. Evaluations of concentration on-task have relied on laboratory methodologies, which encounter difficulties when applied to real work scenarios. Video photoplethysmography (VPPG) can present a solution to these difficulties by extracting physiological changes from videos captured by any conventional camera. Applying machine learning to physiological signals from VPPG can enable contactless detection of task-related concentration. Thirty adults completed a simulated task. Physiological changes were recorded via electrocardiogram (ECG) and VPPG. Pre-trained VGG, support vector machine, and XGBoost were performed on ECG and VPPG signals to detect when participants were on- or off-task. The ensemble method, which combined three machine-learning methods, applied to VPPG signals proved to be highly accurate (∼97%). Among individual machine-learning methods, pre-trained VGG applied to VPPG signals performed the best, comparable to the ensembled method. All analyses showed detection based on VPPG signals to significantly outperform ECG signals. Results establish a proof-of-concept that VPPG and machine learning can be used to detect task-related concentration in a contactless, convenient, and inexpensive fashion. VPPG can enable the detection of task-related concentration in natural work settings.