The TU Delft Safety & Security Institute aims to strengthen the position and increase the impact of Safety & Security research and education by providing seed funding for relevant transdisciplinary collaborative activities at TU Delft.

The following projects have been granted seed funding in the second call for proposals:


1. A real-time decision-making tool for mitigating risks due to heavy localized rain

Dr. F. Fioranelli                 Microwave Sensing Signals & Systems, EWI

Dr. M. Schleiss                 Geoscience and Remote Sensing, CITG

Dr. R. Taormina                Urban Water Infrastructure, CITG

Extreme rain is responsible for millions of damages in the Netherlands every year. A single convective thunderstorm with high peak intensities and strong downdrafts such as the one that occurred on 28/06/2011 near Herwijnen can devastate entire neighbourhoods in a few minutes. As the planet warms up, these type of storms will become more frequent and more dangerous.

To reduce damages and warn users about imminent threats, the Dutch Royal Meteorological Agency KNMI and local water authorities and businesses increasingly rely upon localized/personalized rainfall monitoring and prediction apps (e.g. like However, predictions have limited accuracy, resolution, and update times, which are insufficient for fast-developing, localized phenomena such as microbursts.

At TU Delft, we are working together with KNMI, SkyEcho (weather radar start-up), and HKV (water risk management consultancy), on an innovative concept of hybrid physical statistical rainfall forecasting system based on machine learning. Our group operates a network of diverse weather radars (e.g. PARSAX and MESEWI radars on top of EWI Faculty building, the Rijnmond radar in Rotterdam operated by SkyEcho, portable cloud radars deployed at the Green Village) that we aim to use to enhance and complement the national KNMI network. All these data are integrated into a machine learning framework to achieve more localized and faster forecasts of heavy rainfall (“nowcasting”).

This seed funding will cover implementation costs to realize a small-scale demonstrator on rainfall nowcasting arising from the research of a co-supervised MSc student. The objective is to implement a new visualization and warning tool for risk assessment and decision making in case of heavy rainfall events. The goal of the first demonstrator will be to provide intelligence for real-time decision making and optimization of the water pumping systems for the Delfland area. The longer-term vision is to expand this proof of concept into an automated visualization & decision making tool, and to provide personalized functionalities for TU Delft students such as text warnings to cyclists, motorists or pedestrians with recommendations for how to stay safe in case of incoming severe weather.


2. Environmentally Sustainable Artificial Intelligence in Biomedical Technologies

Dr. Cristina Richie                           Values, Technology and Innovation, TBM

Prof. Patricia Osseweijer                 Biotechnology and Society, TNW

Dr. Lotte Asveld                               Biotechnology and Society, TNW

Prof. Dr. Paddy French                   Bio electronics, EWI

Dr. J. C. Diehl                                 SDI, IO

Prof. Dr. J. Dankelman                   BioMechanical Engineering, 3mE

Dr. Roel Kamerling                         SD, UD

Dr. J.C.J. Wei                                 Medical Instruments & BioInspired Technology, 3mE

Despite identifiable concerns of artificial intelligence (AI) use in healthcare biotechnologies, the most significant ethical issue ought not to be vulnerabilities in the software or potential for exploitation of biodata, but the environmental impact. Healthcare emits a significant amount of carbon in many countries, thus contributing to climate change. In 2017, the Dutch biomedical industry emitted an estimated 15.8 million metric tons of carbon, or 8.1% of the country’s total emissions. While healthcare biotechnologies may be made more sustainable by “greening” the medical lifecycle or by targeting high-impact biotechnologies it would be too laborious to calculate the carbon impact of every aspect of biotech, particularly in the rapidly evolving field of AI. This tension drives my research question: "In absence of comprehensive carbon calculations, how can AI in biotech be more environmentally sustainable?" My proposed project, Environmentally Sustainable Artificial Intelligence in Biomedical Technologies, will organize four collaborative workshops to discuss and develop frameworks for sustainable AI in biomedical technologies.

Funds will cover expenses related to workshops, dissemination of discussion, application of additional funding, and public engagement.


3. Risk of cascading hazards

A transdisciplinary perspective on the impact of consecutive dry and wet spell on flood management

Dr. Elisa Ragno                               Hydraulic Engineering Department, CiTG Gabriela Florentina Nane       Delft Institute of Applied Mathematics, EWI Floortje d’Hont                        Multi-Actor Systems, TBM Oswaldo Morales Napoles     Hydraulic Engineering Department, CiTG

During summer 2020, the longest heatwave in The Netherlands was followed by heavy thunderstorms. Temperature rise and prolonged dry period can undermine the reliability of flood protection systems, exacerbating the risk of flooding when followed by heavy precipitation. Traditional flood management practices consider dry and wet events in isolation leading to a misinterpretation of the risk, especially considering flood management approaches such as “Building with Nature”. Hence, a transdisciplinary approach bridging science (modelling consecutive hazards) and practice (engineering approaches) is fundamental for a robust flood management system ensuring spatial quality and safety of local communities.

The proposal aims to structure the problem of cascading hazards in flood management using transdisciplinary insights and initiating a discussion on cascading hazards to acquire preliminary knowledge, e.g., data, research gaps.

The funding will be used for organizing a 2-day seminar at TU Delft with international experts on cascading hazards; gathering and processing hydro-climatic data for a preliminary analysis of wet and dry spell in the Netherlands and identification of hotspots for further investigation; and disseminating the results.


4. Privacy-aware Robotic Systems

Dr. Laura Ferranti              Cognitive Robotics, 3mE

Dr. Zeki Erkin                    Cyber Security Group, EWI

Mobile robots will soon be part of our daily lives. These technologies have the potential to significantly improve our quality of life, for example, by improving transportation efficiency and safety. Recent surveys, however, showed major societal concerns about mobile robots in terms of security, and privacy. The algorithms used to coordinate the robots play a fundamental role to address these concerns. These algorithms are responsible for the way robots process sensor information, interact with each other, and actuate decisions. From the coordination perspective, there are three key challenges: (i) how to guarantee safety of the robots and humans in complex dynamic environments, (ii) how to guarantee secure operations in the presence of attacks, (iii) how to preserve users’ privacy, which can be compromised by robots sharing information. To address these challenges, our goal is to devise a real-time distributed coordination framework that allows robots to avoid collisions with humans and other robots (safety), while dealing with attacks (security) and preserving user privacy by operating computations over encrypted data (privacy). While the robotics and computer science communities have attempted to address these challenges with ad-hoc solutions, there is an urgent need for generic solutions that can be validated in practice.

This funding will be used to lay the foundation of the collaboration between our teams. We will use the funding to purchase the necessary equipment to test our preliminary algorithms on real robot systems. Our ultimate goal is to submit a joint proposal for the next NWO CyberSecurity call or a KLEIN-2.


5. Research by design for a safe and secure Rijnmond-Drechtsteden delta. M.Z. Voorendt                   Hydraulic Engineering, CiTG M.M Rutten                          Water Management, CiTG J.S. Timmermans                Multi-Actor Systems, TBM

Dr. F. Hooimeijer                          Urbanism, BK

The Dutch Rijnmond-Drechtsteden delta is a highly urbanised area with a high economic value, where the water comes from two sides: from the North Sea and from the rivers. Both are influenced by climate change: through sea level rise and through more extreme high and low river discharges. Economic continuity and spatial development is only possible if flood protection, fresh water supply and climate robustness are assured. Integrated conceptual designs made in interdisciplinary teams are required to explore opportunities, challenges, and solutions in their full complexity. This can be achieved by means of 'research by design'.

The funding be used for hiring an Engineer as a coach for students and coordinator of activities.


6. An Exploratory Study on Process Safety and Asset Integrity Management in the Digital Age

Dr. Ming Yang                                 Safety & Security Science, TBM Pieter van Gelder            Safety & Security Science, TBM Genserik Reniers            Safety & Security Science, TBM Andre de Haan               Chemical Engineering, TNW Anton A. Kiss                  Chemical Engineering, TNW

Ir. Pieter Swinkels                          Chemical Engineering, TNW

Dr. Mihaela Mitici                           Aerospace Transport & Operations, AE

Prof.dr. Maria Nogal                      Integral Design & Management, CiTG

Dr. Xiaoli Jiang                             Transport Engineering & Logistics, 3mE

The extensive use of robotics called for by ambitious programs aiming to redesign industrial production processes, as the Industry 4.0 in the European Union, is dramatically transforming the safety landscape in process industries. With the progression of Industry 4.0, a new generation of process safety and asset integrity management approach is anticipated through the implementation of digital science technologies, such as the machine learning, Internet of Things, big data, cloud computing, smart equipment, and cyber-physical systems, in an integrated process based on condition monitoring data and dynamic risk assessment methods. However, these opportunities come together with new hazards. Under the above context, this project aims to explore the challenges, identify primary research problems, and establish a collaborative research team for greater funding applications. The proposed project also attempts to establish a transdisciplinary collaborative research network among TU Delft researchers and other researchers and industrial practitioners to investigate digitalized process safety and asset integrity management in the process industries.

The funding will be used for relevant data and information gathering, organizing a workshop, and a series of luncheon talks that aim to provide the opportunity for academia and industry to exchange ideas on the challenges, potential solutions, research directions, and develop collaborative research partnerships.


7. EVACUATED: group decision making during evacuations

Dr. Natalie van de Wal                   Multi-Actor Systems, TBM

Dr. Winnie Daamen                       Transport & Planning, CiTG

Dr. Marco Zuniga                           Computer Science, EWI

Dr. Alexei Sharpanskykh                Air Transport & Operations, AE

Dr. Ruggiero Lovreglio                  School of Built Environment, Massey University, New Zealand

Prof Daniel Nilsson                       College of Engineering, University of Canterbury, New Zealand

Prof. Gerta Köster                         Computer Science and Mathematics, Munich University of Applied Sciences, Germany

This project will consist of pilot studies that are aimed to support a VIDI proposal submission in October 2021. The VIDI proposal will take the novel approach of combining state of the art computer modelling with insights from social and cognitive psychology on group-decision making during evacuations to improve speed and survival in emergency evacuations. Specifically, the VIDI project aims to:

A. Determine how groups make decisions during evacuations via lab and field experiments.

B. Propose a new evacuation model for group decision making, based on data gathered in aim A.

C. Validate the evacuation model with real-world data and field experiments.

>The current EVACUATED seed funding project aims to support this VIDI proposal by:

  1. Conducting a literature review to create on an overview of the state of the art of (1) social and cognitive processes of group evacuation decision-making and (2) modelling these processes in evacuation models, supporting aims A and B and the writing of the proposal.
  2. Getting experience with Virtual Reality (VR) research by conducting a replication study of a VR experiment determining if a person follows the crowd (social influence) in route choice behaviour during evacuations, to support lab studies in aim A and the main applicants track record for the VIDI proposal.
  3. Starting to include social attributes in evacuation modelling and validating a new evacuation model, by extracting crowd behaviour patterns using privacy-preserving sensing data (wifi traces or data gathered with mmWAVE sensors), to support modelling efforts in aims B & C, and the main applicants track record for the VIDI proposal.

The funding will be used for effort and hardware, namely payment of student assistants for pilot studies 2 and 3 and VR glasses for pilot study 2 and mmWAVEsensor for pilot study 3.


8. On privacy and security of dynamical networks under partial topology knowledge

Dr. Maksim Kitsak                          QCE, EWI

Dr. Sergio Pequito                          Delft Center for Systems & Control, 3mE

Data breaches associated with security and privacy (S&P) issues are  pervasive in today’s dynamical networks. Yet, we lack the fundamental knowledge on the quantification and mitigation of S&P properties of dynamical networks. Our collaboration seeks to unveil S&P properties under two main premises: (i) S&P of dynamical networks is intertwined with the network’s topology; and (ii) only partial knowledge of the network’s topology is often available. In the long-run, understanding how a network’s topology and dynamics are intertwined will equip policymakers with decision-making mechanisms that enable them to establish regulation on how information is shared and processed locally, but also by third parties that seek to control the network. Ultimately, our results will be a stepping stone to create trustworthy dynamical social networks.

Besides preliminary research investigation, another important activity in this project is to prepare substantial grant applications, such as NWO Open Competition Domain Science – M. Furthermore, we seek to build cooperation with non-academic stakeholders, and boost the TU Delft safety & security institute visability in the context uncertainty quantification and vulnerability assessment, where we bring a new methodology and a set of tools crisscrossing different research fields to assess and quantify S&P properties.


9. SpeakUp! Conversational Agents for Mental Health and Wellbeing

Dr. ir. Ujwal Gadiraju                     Web Information Systems, EWI

Dr. ir. Trivik Verma                        Urban Science and Policy, TBM

Dr. Derek Lomas                          Positive AI, IDE

Our society has constantly dealt with stress induced by rapid social, cultural and technological change. Coping with stress is crucial for a healthy lifestyle, and in turn, affects the safety and wellbeing of society. Studies have highlighted the severe shortage of trained counsellors, professional psychologists, and psychotherapists, to meet the growing demand for mental health support across the world. The COVID-19 outbreak has exacerbated this, affecting the mental health of several people. The student population is particularly vulnerable due to the sudden and radical changes in on-campus education. AI-driven approaches are being proposed for detection and prediction of mental health problems, and for developing solutions. Research in Human-AI collaboration has shown that conversational agents (CAs) can be used to train non-expert individuals in providing effective on-demand therapy. To meet the urgent demand for mental health care providers and improve the accessibility for such support, we propose to use CAs to train non-expert individuals in learning effective counselling techniques (eg. motivational interviewing). We aim to address mental health problems faced by students in universities and people in the broader societal context. To this end, we will organise a joint workshop to identify potential mental health problems that can be tackled by leveraging conversational agents or for developing alternative interventions.

The funding will be used to build a large consortium of relevant stakeholders and organize a workshop to write a larger Horizon Europe proposal.

/* */