Awarded Proposals of Type-A/first Open Call

KITT4SME invited SMEs and midcaps from the AI-developer community with a proposition for the manufacturing industry to join the project and gain access to technical know-how and the chance to create ‘success stories’ for their solutions through the project’s pilot experiments. 46 proposals were submitted to the type-A Open Call, and 10 of them were selected for funding with a total of €966k awarded to the winners. A brief description of each awarded proposal is summarized below including a graphical abstract or a project logo.


AI for Safe and Decent Work (AI4SDW) is an AI solution for human-machine interaction and safety monitoring that works on new or already installed cameras and utilizes the most advanced intelligent edge-AI system powered by next generation AI native chip, that allows real-time processing and assures privacy preservation. At the core of our solution lie a series of proprietary Human Behaviour Understanding AI modules based on people detection, human pose estimation, people segmentation, and tracking. AI4SDW wants to offer the KITT4SME platform an AI-advanced solution to prevent accidents in the manufacturing sector. AI4SDW allows hazard maneuver detection, people counting, area controls, pandemic controls, line crossing, and fall detection assuring continuous monitoring for reaching workers’ zero injury goal.


Accurate prediction of machining time is essential for planning and optimizing production and business processes. Computer Aided Manufacturing solutions estimate job execution times from CNC part programs, but these differ drastically from real processing times. CNCSmart service will predict real execution times of CNC part programs for metal-working industry cutting machines. It will use machine learning techniques to analyse CNC part programs and accurately estimate execution times based on historical data and different cutting parameters, like machine type (e.g., oxyfuel, plasma, laser cutting), material type and thickness, operation (e.g., cutting, marking, drilling, piercing, bevel cutting), and tools handling. CNCSmart will provide a baseline to build prediction services for any CNC-enabled machinery that provides data through OPC/UA, ADS, or other standards.


Emoj is a deep tech startup founded in 2017 to propose advanced technological solutions in the fields of Deep Learning, 3D vision, video analysis and Machine Learning. It offers a series of tools, all based on a proprietary engine for face coding and body detection from video analysis. As part of the KITT4SME funding program EMOJ presents HE.Go.APP. HE.Go.App’s technology can be imagined as the trait d’union between FaMS and IM applications, as it is able to detect the muscular-skeletal risks arising from a dysfunctional working environment, to inform the company’s decision-makers in order to enable maximum optimization of the working environment. The data collected and analyzed through the AI application completes the overall framework of real-time monitoring of the worker[1]factory ecosystem, creating a synergy that has the potential to improve company productivity and the employee health and safety, hence favoring a win-win approach.


OptiPLANT is a cloud-based system that empowers professional and non-professional data scientists to build high-quality ML models for timely and optimised maintenance of the industrial machinery. The vision of OptiPLANT is to allow manufacturing SMEs with limited human and financial resources to set-up and deploy accurate and scalable ML models that will help them optimise the maintenance activities and increase the quality of delivered products. The user uploads the dataset through a web app, trains the models and gets the results. OptiPLANT evaluates a pool of ML models, it automatically selects the best one, and makes results understandable by non-experts. All the complexity is hidden behind a UI that requires zero programming skills.


REORDINIS, an AI-based Reconfiguration Optimization for Resource Distribution and Inventory Integration System addresses the challenge of complex decision-making in production and inventory management. This solution consists of a set of AI technologies, namely simulation & search algorithms and knowledge-based models & queries, which optimize production plan (re)configuration and inventory management decisions to support operators and engineers. REORDINIS stands for a reconfiguration framework for production planning and inventory management enabling production plants to be robust, flexible and modular to handle uncertainties in the market, deviations during production, and systems failures and disruptions.


The THANOS project aims at allowing manufacturing companies to deliver quality products to market faster than their competitors by delivering an AI-based quality control solution in partially automated production processes, that is where parts of the processes are automated and others are not, so that a strong dependence is found between the operators’ and the machines’ performances. This scenario is typical of manufacturing SMEs and mid-caps and will be tackled by adding the human factor to the quality/throughput balance. This will be achieved by identifying throughput levels of relevant nodes in the production process to maintain optimal quality standards, knowing the workers physiological data other than technical data related to the production line. All of the above will be accomplished by integrating one of STAMs’ solutions within the KITT4SME architecture, leveraging on the existing input data and modules’ outputs (i.e FaMS, VIQE, SADS). The THANOS solution analyzes throughput at each production node assessing if it is coherent with the current state of production resources, considering both humans and machinery. This extends the paradigm of the digital factory including human related data from wearable devices allowing a deeper understanding of the production environment, and for this reason, more accurate results can be expected, especially in situations where the human touch is crucial to the quality of the finished products.


AIGreenWaste will develop a solution based on algoWatt AI Green Digesto, an innovative AI solution for waste management and intelligent environmental plants. AI Green Digesto includes decision support tools for biodigester plant operators developed using data and production parameters from the biodigestion plant operating in Nera Montoro in Umbria. A biodigester plant is a system that uses organic waste to produce fertilizers and biogas through an anaerobic process. Current biodigesters must be manually optimized by plant operators to increase biogas production, reduce the amount of residue and remain in the correct range of operating parameters. Though typically equipped with sensors, these systems are not ‘smart machines’ in the sense of Industry 4.0. AI Green will offer the possibility to provide the plant operator with a useful AI decision support for plant management.


The reduction of material and energy consumption is essential for the companies to develop a sustainable business. Due to the importance of reducing the energy consumption and raw materials, numerous research activities and products have been introduced to use emerging technologies such as Big Data processing or Artificial Intelligence to achieve these targets by creating solutions capable of predicting the different consumptions according to the different stages of the production. However, these approaches have some challenges associated with their adoption in some industries, such as injection moulding. The lack of information makes it impossible to create data driven solutions. Thus, the AI4MOS aims to explore the integration of existing injection moulding machines and RAILES to extract data during the process. This data will then be used to create Artificial Intelligence models capable to optimizing the machine allocation according to the predicted energy and materials consumption.


extruAI is a software solution for the extrusion industry. It provides an easy way to support workers in steering extrusion processes. With extruAI, companies can significantly increase the production output and produce high-quality goods – even with low-skilled operators. This is particularly interesting for companies in the pharmaceutical, food and chemical industries. The AI solution is delivered as a ready-to-use software box: A simple integration into existing systems allows it to be used without extensive preparation. Company know-how is not required because we take over the implementation of the machine learning model with our know-how in deep learning and field experience. During the KITT4SME project we will build the solution based on our existing software smartPLAZA.


XtremeStream proposes the development of an embedded Artificial intelligence solution for the early detection and screening of high-risk conditions in industrial environments, especially for gas emissions, volatile organic compounds and other environmental conditions as vibration, high temperatures and long-term high humidity exposure. The solution will be based on the analysis of environmental data from IoT devices installed in subjects’ industrial and work environments and the application of algorithms based on machine learning and artificial intelligence implemented in embedded systems. This solution will contribute greatly to improving the quality of life of users thanks to the early detection of potentially dangerous conditions for their health before they are exposed to them, being able to automatically notify employers. For that reason, this solution is addressing the AI for human-machine interaction, as a mechanism to notify workers and inform them about the environmental conditions in real-time (as a benefit of stream processing with embedded machine learning), at the same time that enables intuitive visualization of propagation maps to understand the risks and impacts.  The system will allow real-time risk detection and evaluation, interacting through the use of alarms, indicators and reports understandable by any end-user.