What is the innovation and the need behind the innovation?: AXIS will apply SotA artificial intelligence to automate the implementation of x-ray inspection systems. AXIS will automate classification software and the hardware specifications.; What is the most important competing solution?: No true competitors exist, a number of domain-specific competitors exist. We know of nobody that use AI for x-ray inspection.; Who will ultimately receive the value created?: Newtec, Qtechnology, and Magnatec all have business plans for the project results. ; What will the applied investment from Innovation Fund Denmark mainly be used for?: Research into a fully automated, x-ray based classification system, including x-ray equipment that works with the AI system.; How will the project results be implemented? Qtechnology and Magnatec will market components and Newtec will market full systems.

ADCendo ApS is a biotech spin-out established in 2017, with a vision to develop new anti-cancer drugs directed at a novel target, in the form of antibody-drug conjugates (ADCs). Current activities revolve around optimization of ADCendos first drug candidate, involving collaboration with international pharma companies. Thereby, this InnoBooster project will enable identification of a lead candidate, and contribute with key progress involving international knowledge- and technology providers.

In 40 years, only one drug, recently failed, has been approved for first-line therapy of sarcoma cancer. Through targeted therapy exploiting a novel antigen, ADCendo seeks to address this unmet need.

We propose cloud-based, personalised treatment to improve real-world adherence and outcomes for type 2 diabetes (T2D) patients. Adherence inhibits most T2D patients from reaching treatment targets. This impacts health, quality of life and cost society. To support care teams and empower patients to self-manage, we integrate data from continuous glucose monitors, connected insulin pens and activity trackers and provide real-time, personalised treatment guidance, data-derived insights, hypoglycemia warnings and dose reminders. Methods include large-scale data collection, development of a high-fidelity T2D model and required algorithms, and clinical feasibility testing of the full solution.

Med Adorno AR vil Adorno Digital tilbyde et produkt, hvormed designere, kunstenere, gallerier og deslignende kan få skabt 3D-modeller af deres værker, hvilket vil øge deres chancer for at sælge værkerne online signifikant. Dette påvises af 2 markedsundersøgelser udarbejdet af Adorno Digital. Adorno AR er et disruptivt produkt idet produktionen er betydeligt mere automatiseret, hvilket gør at produktionen af en 3D-model med Adorno AR er meget billigere og hurtigere, and andre på markedet.

Realize the potential of AI for wind turbine vibration monitoring to boost asset performance and thus accelerate the transition towards a society powered by renewable energy.

The lack of knowledge of the heterogeneous process condition in the large-scale reactors is a major challenge for bioindustries. This project aims to tackle this problem by establishing validated computational simulations with a novel real-time measurement technology (free-floating sensor device by Freesense) for large-scale fermentations. These simulations could be provided to the bioindustry as a valuable tool to design, evaluate, control and optimize various processes.

Projektet vil udvikle analytiske værktøjer som gør det muligt at identificere og kvantificere produkter som dannes ved termisk og katalytisk omdannelse af sukker. Disse værktøjer vil bestå af avancerede kromatografiske separationsteknikker som kan adskille produkterne, i kombination med høj-opløseligt massespektrometri til identifikation. De analytiske metoder vil skabe en bedre forståelse af de kemiske reaktioner som finder sted når sukkermolekyler omdannes i katalytiske processer.

In the description below we refer to the work packages, deliverables, and milestones described in the Work Plan section. An overall timeline of the project is illustrated in Appendix A. The project consists of several partners: Department of Computer Science, University of Copenhagen (from hereon DIKU), who will head the research of the project. Several industry partners (listed previously), who will provide large amounts of support data including tickets, refining the business case and assist with integration and validation/testing. Finally, SupWiz will head the development part of the project and assist with the research. SupWiz will collaborate with the industry partners on data collection and preparation as well as integration and testing. This includes developing technologies for tying everything together in the different channels of support employed by the industry partners (social media, chat bots, e-mail, etc.). The overall structure of the project (described in more detail below) is to have a specification phase, proof of concept phase, agile “mini-projects” of, and a generalization phase. In parallel with this, there will be a research track, which feeds into the mini projects. Each mini-project lasts roughly six months and will work through all three stages of the value chain (from research to testing/validation). The mini projects will use the experience and results from the previous phases as well as the research of the project to collect relevant data, refine the features developed during previous phases, and develop new components. Each mini project ends with integrating and testing in collaboration with the industry partners. The research will feed into the project in the speed results are delivered, and one mini project will contain refinement of the technology based on both new results and experience from previous stages. In this way, partners can be engaged from the beginning and work on integration even before the final results are delivered and implemented. In the generalization track, the goal is to continuously refine the features developed during the project and extract several generic components, which can be further developed, packeted, and marketed as a suite of independent components which can be plugged into existing support systems (such as the systems used by the project partners). SupWiz will be the main responsible party for coordinating between the project partners. This is elaborated further in the project governance section. During the specification phase SupWiz will take special care to account for the diversity of the industry partner involved in the project. This diversity will ensure that the project results in not only customized intelligent support solutions for the involved parties, but also a generic suite of components of independent interest as described above. The research of the project will be split into three work packages with focus on (WP1) question answering, (WP2) ticket routing / expert finding, (WP3) efficient algorithms. Here, WP3 will be focused on refining the methods developed in WP1 and WP2 from an algorithmic perspective. We propose a way to efficiently use deep learning for domain-specific question answering (WP1), covering both factoid and more subjective questions. This is a novel combination, since deep learning requires a lot of training data and is therefore not applicable in small domains, but also there has been very little work, in general, on automatic answering of more subjective questions. We do this in three phases: Knowledge base population, question generation, and question matching. This approach should work well for FAQs, but for tickets there may be several embedded questions and background information, which we will handle using summarization based on different types of parsing. For ticket routing (expert finding, WP2) we propose a two-phase pipeline using online learning, which has not been previously done. In phase 1 we will use advanced data mining to extract expertise indicators for each candidate entity in an organization, and in phase 2 experts will be ranked using a type of online learning called multi-armed bandits. It is the intention to connect the project to the Danish Center for Big Data Analytics Driven Innovation (DABAI) to facilitate maximum synergy with the leading BigData research center in Denmark. The project team is in talks with the DABAI leadership about this. SupWiz is also currently in the process of appointing an advisory board consisting of international experts from both the academic and business world, and is able, through the company’s founders’ and investors’ large networks, to get world-leading experts. In order to accommodate the research of the project, DIKU will hire one 2-year PostDoc and one 1-year student programmer. SupWiz will also hire one full-time resource with a strong theoretical background in algorithms and ML, which assist the remaining SupWiz staff.

Formålet med projektet er at udvikle metoder til at analysere det globale respons samt vækst af skader i vindmøllevinger udsat for forskellige udmattelsestest. Projektet sigter mod at udvikle eksperimentelle metoder til at måle og overvåge vindmøllevinger i under fremtidens fuldskala test, som forventes at blive mere komplicerede for bedre at repræsentere de virkelige laster.Målet er også at udvikle numeriske metoder, som kan bruges til at simulere, hvordan de eksperimentelle målemetoder virker.

Projektets formål er numerisk demonstration og empirisk validering af modeller og strategier for lastreduktion ved hjælp af active flap til anvendelse på store møllevinger og andre hovedkomponenter. Projektet fokuserer på designdrivende lasttilfælde, som har stor industriel relevans specifikt til offshore vindmøller. Last reduktionen har en umiddelbare konsekvens: reduktion af LCOE

For at kunne hjælpe så mange patienter som muligt, er det nødvendigt at nedbringe omkostningerne ved lægemiddelproduktion. Det er dyrt og risikabelt at foretage ændringer i de farmaceutiske produktionsprocesser. En af de bedste måder at mindske risici er udviklingen af effektive forudsigelige procesmodeller. I dette projekt vil vi udvikle nye og innovative modeller ved hjælp af store datasæt med det formål at minimere produktionen af uønskede stoffer ved proces skalaer over 25 kubikmeter.

Hvor nuværende anti-pirat sikring af tankere og containerskibe ligner befæstninger fra 2. verdenskrig, vil Nordreps Advanced Pirate Fence tilbyde en simpel løsning med avanceret teknologi, der vil gøre det både let og effektivt at sikre skibe mod kapring. Løsningen reducerer omkostningerne til installation af værnemidler med en faktor 10, samtidig med at risikoen for skader på mandskab og materiel under installation reduceres til en minimum.

It is the project´s aim to create a fast, reliable closed-loop control system for Tasso´s continuous casting unit that will increase the company´s product quality and production rate. Such systems do not exist today and will have to be developed from a thorough experimental analysis combined with application of numerical modelling of the casting process. The project should lead to a first version of a closed-loop control system.

Efficacy of KP405 has been extensively characterized in multiple animal models of neurodegeneration. This proposal has been submitted with the purpose to advance the development of KP405 to human trials. Specifically, this proposal entails completion of the manufacturing and toxicity studies of KP405 in animals, required for regulatory submission of an investigational new drug (IND)/clinical trial application (CTA) and subsequent evaluation of safety, tolerability, and efficacy in humans.