The main impetus for the global efforts toward the current digital transformation in almost all areas of our daily lives is due to the great successes of artificial intelligence (AI), and in particular, the workhorse of AI, statistical machine learning (ML). The intelligent analysis, modeling, and management of agricultural and forest ecosystems, and of the use and protection of soils, already play important roles in securing our planet for future generations and will become irreplaceable in the future. Technical solutions must encompass the entire agricultural and forestry value chain. The process of digital transformation is supported by cyber-physical systems enabled by advances in ML, the availability of big data and increasing computing power. For certain tasks, algorithms today achieve performances that exceed human levels. The challenge is to use multimodal information fusion, i.e., to integrate data from different sources (sensor data, images, *omics), and explain to an expert why a certain result was achieved. However, ML models often react to even small changes, and disturbances can have dramatic effects on their results. Therefore, the use of AI in areas that matter to human life (agriculture, forestry, climate, health, etc.) has led to an increased need for trustworthy AI with two main components: explainability and robustness. One step toward making AI more robust is to leverage expert knowledge. For example, a farmer/forester in the loop can often bring in experience and conceptual understanding to the AI pipeline—no AI can do this. Consequently, human-centered AI (HCAI) is a combination of “artificial intelligence” and “natural intelligence” to empower, amplify, and augment human performance, rather than replace people. To achieve practical success of HCAI in agriculture and forestry, this article identifies three important frontier research areas: (1) intelligent information fusion; (2) robotics and embodied intelligence; and (3) augmentation, explanation, and verification for trusted decision support. This goal will also require an agile, human-centered design approach for three generations (G). G1: Enabling easily realizable applications through immediate deployment of existing technology. G2: Medium-term modification of existing technology. G3: Advanced adaptation and evolution beyond state-of-the-art.

Smart farming (SF) applications rely on robust and accurate computer vision systems. An important computer vision task in agriculture is semantic segmentation, which aims to classify each pixel of an image and can be used for selective weed removal. State-of-the-art implementations use convolutional neural networks (CNN) that are trained on large image datasets. In agriculture, publicly available RGB image datasets are scarce and often lack detailed ground-truth information. In contrast to agriculture, other research areas feature RGB-D datasets that combine color (RGB) with additional distance (D) information. Such results show that including distance as an additional modality can improve model performance further. Therefore, we introduce WE3DS as the first RGB-D image dataset for multi-class plant species semantic segmentation in crop farming. It contains 2568 RGB-D images (color image and distance map) and corresponding hand-annotated ground-truth masks. Images were taken under natural light conditions using an RGB-D sensor consisting of two RGB cameras in a stereo setup. Further, we provide a benchmark for RGB-D semantic segmentation on the WE3DS dataset and compare it with a solely RGB-based model. Our trained models achieve up to 70.7% mean Intersection over Union (mIoU) for discriminating between soil, seven crop species, and ten weed species. Finally, our work confirms the finding that additional distance information improves segmentation quality.

Near infrared (NIR) spectroscopy is a potentially valuable tool for estimating the process parameters of anaerobic digestion (AD) in agricultural biogas plants. In addition to precision and accuracy, the evaluation of model robustness versus changes in the feedstock composition and process stages are needed to implement this analytical technology into common practice. This paper reports the first step of a global modelling approach, addressing the need for increased calibration robustness for the estimation of the process parameters volatile solids (VS), ammonium (NH₄–N), total inorganic carbon (TIC), total volatile fatty acids (VFA), acetic acid and propionic acid in the fresh matter (FM) of digester sludge. Spectra from samples in different training sets, varying with respect to their origin and feedstock composition, were assessed using partial least square regression. The comparison of the offline calibration results among the training sets revealed that an increase in the heterogeneity of the sample matrices did not result in a relevant performance loss of the NIR models. With a root mean square error of cross-validation ( RMSECV) of 4.0 g kg⁻¹ and 0.16 g kg⁻¹ FM, VS and NH₄–N exhibited the highest potential for estimation via NIR spectroscopy. While the strong X–Y relationship for the structurally NIR-inactive TIC ( RMSECV = 0.80 g kg⁻¹ FM) indicated a satisfactory screening potential, further study is required before application. The volatile fatty acid parameters, which are useful for detecting short-term changes of the AD, did not result in good NIR models. However, there appears to be a realistic potential for a global VFA model with an estimation error of 0.9 g kg⁻¹ FM, which may support the use of NIR-based rapid screening of the dynamics of the acidity level in digester sludge.

Biogas plants can contribute to future energy systems’ stability through flexible power generation. To provide power flexibly, a demand-oriented biogas supply is necessary, which may be ensured by applying flexible feeding strategies. In this study, the impacts of applying three different feeding strategies (1x, 3x and 9x feeding per day) on the biogas and methane production and process stability parameters were determined for a biogas plant with a focus on waste treatment. Two feedstocks that differed in (1) high fat and (2) higher carbohydrate content were investigated during semi-continuous fermentation tests. Measurements of the short chain fatty acids concentration, pH value, TVA/TIC ratio and total ammonium and ammonia content along with a molecular biology analysis were conducted to assess the effects on process stability. The results show that flexible biogas production can be obtained without negative impacts on the process performance and that production peaks in biogas and methane can be significantly shifted to another time by changing feeding intervals. Implementing the fermentation tests’ results into a biogas plant simulation model and an assessment of power generation scenarios focusing on peak-time power generation revealed a considerable reduction potential for the needed biogas storage capacity of up to 73.7%.

Reforestation is a threat to permanent grasslands in many alpine regions. Using these areas to produce biogas energy may help to preserve these important landscapes and save fossil fuels by adding a renewable local heat and electricity source. This case study compares (a) a status quo (SQ) reference scenario with heating oil, wood-chips, and grid electricity as municipal energy sources, and (b) a hypothetical local biogas (LB) scenario (to also be used as a municipal energy source) based on a 500-kW_el biogas plant with steam explosion pretreatment. Here, hay from previously unused grassland is the main biogas substrate, whereas, in the reference SQ scenario, these grasslands remain unused. Life cycle assessment (LCA) results for LB and SQ scenarios are significantly different at p < 0.05 in all six impact categories. In three categories, the LB scenario has lower impacts than the SQ scenario, including climate change (0.367 CO₂-eq kWhel-1 versus 0.501 CO₂-eq kWhel-1). Dominant contributions to climate change in the SQ scenario are from the extant municipal energy sources that the LB biogas plant would replace; in the LB scenario, important contributions include unburned methane from the biogas plant, as well as CO₂ emissions from hay production machines. In summary, important environmental impacts can be reduced and alpine grasslands can be preserved by biogas production from that grass. The advantages of integrating a local biogas plant in municipal energy and waste systems depend strongly on the extant municipal energy system characteristics.

Grassland vegetation typically comprises the species groups grasses, herbs, and legumes. These species groups provide different functional traits and feed values. Therefore, knowledge of the botanical composition of grasslands can enable improved site-specific management and livestock feeding. A systematic approach was developed to analyze vegetation of managed permanent grassland using hyperspectral imaging in a laboratory setting. In the first step, hyperspectral images of typical grassland plants were recorded, annotated, and classified according to species group and plant parts, that is, flowers, leaves, and stems. In the second step, three different machine learning model types—multilayer perceptron (MLP), random forest (RF), and partial least squares discriminant analysis (PLS-DA)—were trained with pixel-wise spectral information to discriminate different species groups and plant parts in individual models. The influence of radiometric data calibration and specific data preprocessing steps on the overall model performance was also investigated. While the influence of proper radiometric calibration was negligible in our setting, specific preprocessing variants, including smoothening and derivation of the spectrum, were found to be beneficial for classification accuracy. Compared to extensively preprocessed data, raw spectral data yielded no statistically decreased performance in most cases. Overall, the MLP models outperformed the PLS-DA and RF models and reached cross-validation accuracies of 96.8% for species group and 88.6% for plant part classification. The obtained insights provide an essential basis for future data acquisition and data analysis of grassland vegetation.