Six application-oriented projects can start with funding from the Open Technology Programme from the NWO domain Applied and Engineering Sciences (AES). Two of these six proposals, and one as co-applicant, have been awarded to Delft researchers. In early June, this OTP funding was awarded and in July these two researchers received their OTP funding. NWO is supporting these application-oriented projects with more than 5.5 million euros. Industry partners and other organisations are contributing another amount of over 885.880 euros.
The three Delft honoured projects are:
TINA: Accelerating non-NN algorithms using high-performance NN hardware accelerators
Dr. ir. Zaid Al-Ars, Electrical Engineering, Mathematics & Computer Science (EEMCS)
This project aims to develop a novel framework for implementing non neural network (NN) algorithms on NN accelerators such as GPUs, TPUs or FPGAs. This approach maps mathematical and logic functions as a series of convolutional and fully connected layers. This way, it becomes possible to execute non-NN algorithms on NN hardware accelerators efficiently, and ensure their portability to any NN accelerator platform, which allows for fast design cycles. Our main focus is on industrially relevant signal processing algorithms to address computational challenges in the signal processing and scientific computing fields in a collaboration between TUDelft, Erasmus MC and ASTRON.
Read more in the and at the researchpage of Zaid.
POINT-TWINS: Point clOud-based dynamIc coNTinuous-scale adapTive urban WINd Simulation
Dr. Azarakhsh Rafiee, Architecture and the Built Environment (BK)
Co-applicant(s): dr. ir. Frits de Prenter (Aerospace Engineering) & ir. Peter van Oosterom (Architecture and the Built Environment)
Urbanization growth affects wind patterns, impacting air pollution, thermal comfort, health and energy efficiency. For optimal design and management of urban areas we need accurate wind simulations. Current wind simulations are computationally demanding, limiting their use on large areas. For areas including complex geometries (e.g. urban areas including vegetation) mesh preparation is challenging and mesh quality has considerable impact on simulation outcome. We will develop time-scale-space adaptive meshfree method for urban wind simulation utilizing Dutch Airborne Laser Scanning open data. Given high detail in point clouds and adaptive nature of point distribution, we expect the method be superior in quality.
Take a look at the staffpages of Azarakhsh , Frits and Peter.
Nanophotonically enhanced absorption in thin silicon pixels for improved image sensors (AHEAD)
W.L. Vos, Twente of University
Co-applicant(s): dr. Paolo Sberna (Electrical Engineering, Mathematics & Computer Science) and dr. M.J. Goodwin & A. Lagendijk (Twente University)
To record optical images like specialized images to selfies for further processing, the relevant amount of light is absorbed by an image sensor. Such a sensor consists of many sensitive elements -pixels - made of silicon with modern semiconductor nanotechnology. Since nanotechnology is continuously miniaturizing and silicon absorbs light weakly, absorption must be greatly increased to convert light much more efficiently into charge that is detected as a digital image. The innovation of AHEAD is to increase absorption by trapping light in pixels with smart photonic bandgap nanostructures. AHEAD is so flexible that different bands of light can be optimized.
Read more in the Faculty News release of EEMCS and at the researchpage of Paolo.
The provides funding for application-oriented technical-scientific research that is free and unrestricted and is not hindered by disciplinary boundaries. The programme offers companies and other organisations an accessible way to participate in scientific research that is intended to lead to societal and/or scientific impact
Please read the
