VTT Research System
Not a member yet
161399 research outputs found
Sort by
Towards Sustainable Weed Management Using Lightweight Deep Learning Model
No full textThe exponential growth of population has resulted in food safety becoming a major concern in global context. To provide food for people and livestock worldwide, it is crucial to implement intelligent solutions that cater to the specific needs of crop cultivation, while maintaining soil quality. Maize holds higher potential than other major crops as it is widely used as industrial raw material, bio-ethanol production, feed and fodder for cattle, besides its primary use as food. Weed management plays a crucial role in maize agricultural practices as it helps ensure optimal crop growth and yield. Conventional weed control methods have limitations that hinder their effectiveness for future weed management. Also, Weed management has become increasingly challenging due to the over-reliance on herbicides that has accelerated the evolution of herbicide-resistant weeds among increasing concerns about effect of pesticides on environment and human health. As a result, there is a growing need for an integrated approach that combines different strategies and utilizes new technologies towards precise and efficient weed management. The work in the following paper utilizes the YOLOv5 object detection algorithm to detect and classify weeds in images. The trained model can then be used for inference on new images to identify and classify weeds.</p
Enhancing Safety in Autonomous Vehicles Using Advanced Deep Learning-Based Pothole Detection
No full text Autonomous vehicles possess the potential to revolutionize transportation by significantly enhancing safety and efficiency. However, their success hinges on overcoming numerous challenges including the detection of potholes which pose significant risks to vehicles and passengers. Consequently the identification and remediation of these obstacles are crucial for the safety of autonomous systems. This research introduces YOLO v8 as a formidable solution for pothole detection predicated on the latest You Only Look Once (YOLO) algorithm. Utilizing deep learning techniques this system identifies potholes in real-time enabling autonomous vehicles to circumvent potential hazards and diminish the risk of accidents. Extensive testing with publicly accessible datasets reveal that this approach surpasses contemporary state-of-the-art methodologies in both precision and speed. Various data augmentation strategies are also examined to further enhance detection performance. Empirical evidence indicates that the YOLO v8-based pothole detection system exhibits superior efficacy compared to other analogous systems. This advancement signifies that autonomous driving can be rendered safer and more reliable marking a pivotal milestone in the enhancement of road safety.</p
Large-scale forest resource mapping with spatial gaps in the training data:Comparison of different modeling approaches
No full textForest attribute maps are essential for supporting local decision-making regarding forest resource use. Such maps are produced by combining remote sensing and field data through various modeling approaches. When mapping across large areas, spatial gaps in field data used for model training are common. Our study evaluates the performance of three methods—k-Nearest Neighbor (k-NN), Random Forests (RF), and Multi-Layer Perceptron (MLP)—for forest resource mapping across Norway, Sweden, and Finland in an experimental setup with respect to availability of field data around the target area. Models were trained with sample plot sizes (N) ranging from 100 to 3000. RF consistently produced the most accurate predictions in terms of relative bias and RMSE. While spatial gaps in the training data (radius: 7–141 km) affected %RMSE of broad-leaved above ground biomass (AGB), they had minimal impact on %RMSE of both local and country-level predictions of total AGB and volume. For RF with N=3000, %RMSE of total AGB ranged between 53%–55% in Finland and Sweden, and 70%–72% in Norway across gap sizes. However, %bias increased for local predictions across the whole study region with larger gaps: RF with N=500 showed bias of −12%–12% (7 km gap) and −17%–28% (78 km gap). Similarly, country-level %bias of total AGB for Norway increased from −1.7% to −3.7% with larger gaps. In conclusion, spatial gaps in training data can significantly affect bias in predictions. Therefore, forest attribute maps should always be accompanied by metadata describing the training data used.</p
Production of high-concentration CO2 from electrified limestone calcination for carbon capture applications
Full text linkOperating electrically heated kilns under high-CO2 atmospheres can increase CO2 capture efficiency but creates reducing conditions that drive CO formation. In this work, CO generation during limestone calcination in a 280 kW electrically heated rotary kiln at 75 vol-% CO2 and low O2 concentration is investigated. Equilibrium calculations indicate that sulphide and sulphite phases in limestone decompose, releasing SO2 and promoting CO formation. Complementary packed-bed experiments confirm that sulphur species are a major CO promoter and reveal a synergistic interaction between sulphur compounds and elevated CO2 levels. Using low-sulphur limestone could suppress CO emissions. Where low-sulphur feedstocks are unavailable, targeted electrolytic O2 or air injection coupled with indirect limestone preheating is proposed to strip sulphur and preserve the high-purity CO2 stream which will improve the efficiency of electrified kilns integrated with a carbon capture process
Boosting softwood hemicellulose hydrolysis:Enzymes from a new fungi Penicillium rotoruae remarkably improve CTec-2 hydrolysis efficiency and reduce sugar production costs
No full textEconomic production of fermentable sugars from lignocellulosic biomass is critical for the biorefinery applications in the bioeconomy industry. This study demonstrates effective enzymatic hydrolysis of recalcitrant softwood using newly identified fungus Penicillium rotoruae. Initially, nineteen fungal isolates were screened on softwood galactoglucomannan (GGM), with nine showing strong responses in the liquid culture. Trichoderma viride, Penicillium rotoruae, and Amorphotheca resinae showed highest β-mannanase, β-mannosidase, and α-galactosidase activities. P. rotoruae demonstrated superior main chain cleaving enzyme activities, while A. resinae excelled in the side chain cleaving activity. The crude enzyme of P. rotoruae was evaluated on two Pinus radiata substrates. Using soluble GGM, P. rotoruae released 34.3 % monomeric sugars (32.1 g/L reducing sugars), outperforming commercial CTec-2 (22.9 % and 23.2 g/L respectively). Co-application of CTec-2 with P. rotoruae enzymes increased monomeric sugar yield to 56.3 %, with galactose, mannose, and glucose increasing 20-, 3.6-, and 2.2-fold respectively. Using insoluble pulp, co-application yielded 88 % of monomeric sugars (20.2 g/L reducing sugars) representing an increase of 20 % soluble sugars relative to CTec-2 used alone. Techno-economic analysis indicated an increase in annual EBITDA, a positive ROCE and sugar cost savings of NZD 125/t demonstrating significant economic potential for softwood biorefineries.</p
Digitalisation in geosciences for environmental protection
No full textData Science (Digitalization and Artificial Intelligence) became more than an important facilitator in various domains in fundamental and applied sciences as well as industry and is disrupting the way of research already to a large extent. Originally, data sciences were viewed to be well-suited, especially, for data-intensive applications such as image processing, pattern recognition, etc. In the recent past, particularly, data-driven and physics-inspired machine learning methods have been developed to an extent that they accelerate numerical simulations and became directly applied in the nuclear waste management cycle. In addition to process-based approaches for creating surrogate models, other disciplines such as virtual reality methods and high-performance computing are leveraging the potential of data sciences more and more. The present challenge is utilizing of the best experimental and monitoring data as well as model concepts and tools to integrate multi-chemical-physical, coupled processes, multi-scale and probabilistic simulations in Digital Twins (DT) able to mirror or predict the performance of its corresponding existing or future physical implementations including workflows. The call for the Topical Collection was initiated from different actors, including research entities, technical support organizations and nuclear waste management organizations of the European projects EURAD (European Joint Programme on Radioactive Waste Management) and PREDIS (Pre-disposal Management of Radioactive Waste). The Topical Collection attracted a large number of manuscripts, more than eighty of which were published. These articles reveal a strong academic focus on using machine learning to map and assess soil and groundwater resources, hydrology and land use, landslides, and climate protection. They also highlight the core theme of nuclear waste management.</p
Impact pathways:geo-operations for turning plastic waste into carbon capture
No full textPurpose – We explore how an operations and supply chain approach to geo-engineering can enhance circular economy approaches and mitigate climate change. We illustrate how such geo-operations – specifically the combination of plastics and biowaste processing – can be systematically leveraged for carbon capture. Design/methodology/approach – The study applies production theory and operations management perspectives to develop a carbon transfer model. It traces carbon flows through the extended plastics supply chain and interconnected natural systems, from raw material inputs, through production and reuse cycles, to the ultimate disposal. By mapping carbon transfers between natural systems and artificial systems, the framework highlights the systemic impact pathways for operations and supply chain management. Findings – Single interventions such as bio-based materials, chemical recycling or policy instruments have limited impact in isolation. However, when combined systemically, these individual solutions can form geo-engineering operational pathways that draw out atmospheric carbon and refossilize it, thus transforming the plastics technosphere from a source of emissions to a means for carbon capture. Research limitations/implications – The study is conceptual and develops theoretical propositions on systemic impact, rather than presenting empirical findings. Future research should empirically investigate the feasibility, scale and trade-offs of the proposed geo-operations pathways. Practical implications – The carbon transfer model and impact pathways guide policymakers, producers and waste managers on integrating the circular economy and geo-operations for climate change mitigation and carbon capture. Social implications – By reframing plastics not only as a source of problematic waste but also as a possible vehicle for climate mitigation, the paper suggests new opportunities and responsibilities for industry and society. Originality/value – This paper proposes the development of geo-operations as a systemic pathway for integrating circular economy and carbon sequestration interventions. It also presents a framework to assess the impact of combinations of interventions on carbon flows.</p
OpenTextile-NIR: Near-infrared hyperspectral imaging and photography dataset for optical identification of textiles
No full textSee description at dataset page in Zenodo: https://zenodo.org/records/1826917
Production of high-concentration CO2 from electrified limestone calcination for carbon capture applications
Full text linkOperating electrically heated kilns under high-CO2 atmospheres can increase CO2 capture efficiency but creates reducing conditions that drive CO formation. In this work, CO generation during limestone calcination in a 280 kW electrically heated rotary kiln at 75 vol-% CO2 and low O2 concentration is investigated. Equilibrium calculations indicate that sulphide and sulphite phases in limestone decompose, releasing SO2 and promoting CO formation. Complementary packed-bed experiments confirm that sulphur species are a major CO promoter and reveal a synergistic interaction between sulphur compounds and elevated CO2 levels. Using low-sulphur limestone could suppress CO emissions. Where low-sulphur feedstocks are unavailable, targeted electrolytic O2 or air injection coupled with indirect limestone preheating is proposed to strip sulphur and preserve the high-purity CO2 stream which will improve the efficiency of electrified kilns integrated with a carbon capture process
Large Eddy Simulation of environmental impacts on mass transport in laboratory-scale vertical farm
No full textThe impact of environmental factors on airflow and mass transport within a laboratory-scale vertical farm is investigated using Computational Fluid Dynamics. Large Eddy Simulation models complex airflow behaviour, while solving enthalpy and mass transport equations yields temperature, humidity, and CO2 concentration. The Eulerian-Lagrangian approach simulates the free-fall of water droplets in the dehumidifier-cooling system. Humidity and CO2 consumption/production by plants and utilities are modelled as volumetric sources/sinks. An experimental campaign is conducted to measure temperature, relative humidity, and CO2 above cultivation beds, validating the numerical setup with mean absolute errors of 0.8%, 2.2%, and 3.9%, respectively. Analysing the airflow shows that the free fall of droplets is the dominant mechanism driving airflow characteristics. We investigate the effects of wall confinement, number of lamps, and location of lamps on the mass transport. Curtains were used to divide each cultivation bed into three regions to assess the wall confinement effect. Results show the overall adverse effect of curtains on mass transport. In more detail, mass transport is enhanced when the curtains and streamlines are aligned parallel, whereas it is reduced when they are perpendicular. Increasing the number of operative lamps improves the uniformity of mass distribution on the upper cultivation beds due to a stronger positive buoyancy. Positioning lamp-induced buoyant flow within the droplet’s lateral momentum injection zone further enhances vertical mass transport. These findings highlight the impact of environmental factors on mass transport, offering insights for more efficient designs of indoor vertical farms.</p