Endovascular interventions are an established class of procedures within minimally invasive surgery (MIS). They enable the treatment of cardiovascular diseases through small incisions in the body by using flexible instruments. Conventionally, these instruments are manually operated, which restricts their precision, and limits their applicability. The magnetic actuation of instruments for endovascular interventions creates a novel and effective steering alternative. Even in deeply seated regions, magnetic flexible instruments provide clinicians with dexterity, while retaining minimal access. Thereby, they permit a range of advanced surgical tasks unattainable otherwise. Nevertheless, to be remotely actuated, such instruments rely on magnetic fields originating from outside the body. Thus, the aim of RAMSES is to develop and evaluate a clinic-ready robotic system capable of generating external fields during endovascular interventions. The RAMSES system will become an enabling technology for the clinical use of magnetic surgical instruments. It will truly revolutionize MIS, opening a new market for advanced diagnosis and treatment options. The RAMSES system will contain optimized electromagnetic actuators, located on robotic manipulators and powered by dedicated control software. The resulting versatile clinical framework will be applicable to a wide range of surgical instruments. This includes both commercially-available magnetic catheters as well as novel experimental designs. As a consequence, RAMSES will satisfy the needs of clinicians to further expand the effectiveness and availability of MIS techniques. It will provide an indispensable clinical tool for accurate and comprehensive surgical interventions in hard-to-reach locations within the human body. RAMSES aspires to turn magnetic actuation into a commercially successful technology. It involves strong industrial collaborations and a dedicated business development team in an ambitious quest to make that happen.

Traffic safety is the fundamental criterion for vehicular environments and many artificial intelligence-based systems like self-driving cars. There are places, e.g., intersections and shared spaces, in the urban environment with high risks where vehicles and vulnerable road users (VRUs) such as pedestrians and cyclists directly interact with each other. By advancing starte-of-the-art artificial intelligence methodologies, this project VeVuSafety aims to build a privacy-aware deep learning framework to learn road users’ behaviour in various mixed traffic situations for the safety between vehicles and VRUs. VeVuSafety proposes a 3D environment model based on 3D point cloud for privacy protection — private information like license plates and face is anonymized. Then, within this environment model, an end-to-end deep learning framework using camera data will be built for multimodal trajectory prediction, anomaly detection, and potential risk classification based on deep generative models such as Variational Auto-Encoder. Additionally, an active privacy mechanism will also be adopted by application of the differential privacy mechanism to help the deep learning models prevent model-inversion attack. Moreover, the framework’s generalizability will be investigated by exploring the Normalizing Flows approach for domain adaption. The framework’s performance will be validated at different intersections and shared spaces using real-world traffic data. Besides road user safety and privacy, VeVuSafety can help traffic engineers and city planners to better estimate the design of traffic facilities in order to achieve a road-user-friendly urban traffic environment. Furthermore, the success of VeVuSafety will enhance the fellow’s scientific knowledge and project management skills to become an artificial intelligence expert for traffic safety and Intelligent Transportation Systems.

Melting and dissolution induce temperature and concentration gradients in liquid systems. These gradients induce flows, namely buoyancy driven flows on large scales and phoretic flows on small scales. Such flows locally enhance or delay the melting or dissolution process and thus determine the objects’ shape. On large scales, a relevant example for the climate are glaciers and icebergs melting into the ocean, where cold and fresh meltwater experiences buoyant forces against the surrounding ocean water, leading to flow instabilities, thus shaping the ice and determining its melting rate. Another example is the dissolution of liquid CO2 in brine for CO2 sequestration. Next to buoyant forces also phoretic forces along the interfaces come into play. For dissolving drops at the microscale the phoretic forces become dominant. The resulting Marangoni flow not only affects their dissolution rate, but can also lead to their autochemotactic motion, deformation, or even splitting. In spite of the relevance for these and many other applications, such multicomponent, multiphase systems with melting or dissolution phase transitions are poorly understood, due to their complexity, multiway coupling, feedback mechanisms, memory effects & collective phenomena. The objective of this project is a true scientific breakthrough: We want to come to a quantitative understanding of melting & dissolution processes in multicomponent, multiphase systems, across all scales and on a fundamental level. To achieve this, we perform a number of key controlled experiments & numerical simulations for idealized setups on various length scales, inspired by above sketched problems, but allowing for a one-to-one comparison between experiments and numerics/theory. For the first time, we will perform local measurements of velocity, salt concentration, and temperature and connect them to global transport processes, to arrive at a fundamental understanding of such Stefan problems in multicomponent systems.

Virtually all aspects of society, industry and science are significantly impacted by increasingly complex computers and the software that they run. A major objective within computer science is to ensure that these computer systems are formally correct by developing ways to prove that a system correctly implements certain given properties, or ways to construct such a system. In this domain of computer science, an important topic is parity games. Many real-world systems run continuously and properties for such systems are described in temporal logics such as linear temporal logic (LTL) and various logics derived from or related to LTL. Solving parity games computes whether a given system has a property specified in these temporal logics, or constructs a system that implements such a property. Parity games are in addition a compelling subject for theoretical computer science because they are believed to be in the complexity class P; however, this has been an open question for over 20 years. In recent years, new algorithms have been discovered that solve parity games in quasi-polynomial time, while at the same time researchers have found quasi-polynomial lower bounds for several families of algorithms.In my publication on tangle learning, I propose the notion of a tangle and show that existing parity game solving algorithm implicitly explore tangles when reasoning about how a player can force the opponent to move through the parity game from A to B. The aim of this proposal is to understand the inherent complexity of solving parity games algorithmically. The strategy to achieve this is to obtain fundamental insights into the structure of tangles, how they are handled by parity game solving algorithms and how they arise in practical games.

Diagnostic agents are currently injected into the body in an uncontrolled way and visualized using non-real-time imaging modalities. Delivering agents close to the organ and magnetically guiding them to the target would permit a myriad of novel diagnostic and therapeutic options, including on-site pathology and targeted drug delivery. Such an advance would truly revolutionize minimally invasive surgery (MIS). Presently MIS often involves manual percutaneous insertion of rigid needles. These needles deviate from their intended paths due to tissue deformation and physiological processes. Inaccurate needle placement may result in misdiagnosis or ineffective treatment. Thus, the goal of ROBOTAR is to design a robotic system to accurately steer flexible needles through tissue, and enable precise delivery of agents by magnetically guiding them to a designated target. There are several challenges: 3D models describing the evolving needle shape are not available, real-time control of flexible needles using 3D ultrasound (US) images has not been demonstrated, and US-guided tracking of magnetic agents has not been attempted. These challenges will be overcome by using non-invasively (via US) acquired tissue properties to develop patient-specific biomechanical models that predict needle paths for pre-operative plans. Intra-operative control of flexible needles with actuated tips will be accomplished by integrating plans with data from US images and optical sensors. Ultrafast US tracking methods will be coupled to an electromagnetic system to robustly control the agents. A prototype will be evaluated using microrobots and clusters of nanoparticles in scenarios with realistic physiological functionalities. The knowledge gained will be applicable to a range of flexible instruments, and to an assortment of personalized treatment scenarios. This research is motivated by the existing need to further reduce invasiveness of MIS, minimize patient trauma, and improve clinical outcomes.