AIT Austrian Institute of Technology

Computer vision is vital for various applications like object tracking for autonomous driving or quality assurance. Hence, assuring that computer vision fulfills given quality criteria is essential and requires sufficient testing. In previous work, authors introduced a testing method relying on image modifications for a photometric stereo application. Image modifications include pixel errors or the rotation of images to be analyzed, revealing a substantial impact on the computed outcome of the photometric stereo application, depending on the applied modification. This paper focuses on whether we can reproduce the impact of image modifications in a real-world setup. In particular, we compare the impact of the rotation of the analyzed sample with the rotation modification applied to the image of the sample. The comparison indicates a similar effect when using rotation, showing that testing based on image modifications is valuable for verifying computer vision applications.

Detecting cybersecurity vulnerabilities in Unmanned Aerial Systems (UAS) is essential to ensure the safe operation of drones. This supports the determination of cybersecurity objectives and the description of security requirements needed to achieve these objectives. However, it is challenging to automate this process to identify potential cyber threats and ensure the correctness of the applied security requirements, especially in a complex system such as a UAS network. In this work, we use ThreatGet as a threat modelling tool to identify potential cyber threats in UAS and highlight existing security vulnerabilities. This assists in determining the appropriate security requirements that could be implemented to achieve our security goal. We then develop a novel ontology-based threat modelling approach to infer a set of security threats based on the applied security requirements and then check the effectiveness of these requirements against threats to ensure these requirements are fulfilled.

Current automotive safety standards are cautious when it comes to utilizing deep neural networks in safety-critical scenarios due to concerns regarding robustness to noise, domain drift, and uncertainty quantification. In this paper, we propose a scenario where a neural network adjusts the automated driving style to reduce user stress. In this scenario, only certain actions are safety-critical, allowing for greater control over the model’s behavior. To demonstrate how safety can be addressed, we propose a mechanism based on robustness quantification and a fallback plan. This approach enables the model to minimize user stress in safe conditions while avoiding unsafe actions in uncertain scenarios. By exploring this use case, we hope to inspire discussions around identifying safety-critical scenarios and approaches where neural networks can be safely utilized. We see this also as a potential contribution to the development of new standards and best practices for the usage of AI in safety-critical scenarios. The work done here is a result of the TEACHING project, an European research project around the safe, secure and trustworthy usage of AI.

Critical infrastructures are making increasing use of digital technology for process control. While there are benefits, such as increased efficiency and new functionality, digitalization also introduces the risk of cyber-attacks to systems that support critical functions. A valuable target in these Industrial Control Systems (ICSs) are the Programmable Logic Controllers (PLCs) controlling the machinery that manage a physical process. PLCs have proven to be vulnerable to a range of cyber-attacks in the past; however, newer technologies such as embedded servers and virtualization have the potential to improve this situation and be used to monitor a PLC’s function. In this article, the implementation of a Host-based Intrusion Detection System (HIDS) for a modern PLC is described. This method uniquely makes use of native technologies on the PLC to monitor a dynamic simulated process in real time. Both the PLC’s integrity (checksum, file size, etc.) and the process control are monitored to determine whether the PLC has been compromised in a cyber-attack. The proposed solution detects a range of attacks, even when the PLC’s control logic is compromised, and – unlike previous PLC HIDS methods – requires no modification of the underlying PLC technology.

Formal specifications are essential to express precisely systems, but they are often difficult to define or unavailable. Specification mining aims to automatically infer specifications from system executions. The existing literature mainly focuses on learning properties defined on single system executions. However, many system characteristics, such as security policies and robustness, require relating two or more executions, and hence cannot be captured by properties. Hyperproperties address this limitation by allowing simultaneous reasoning about multiple executions with quantification over system traces. In this paper, we propose an effective approach for mining Hyper Signal Temporal Logic (HyperSTL) specifications. Our approach is based on the syntax-guided synthesis framework and allows users to control the amount of prior knowledge embedded in the mining procedure. To the best of our knowledge, this is the first mining method for hyperproperties that does not require a pre-defined template as input and allows for quantifier alternation. We implemented our approach and demonstrated its applicability and versatility in several case studies where we showed that we can use the same method to mine specifications both with and without templates, but also to infer subsets of HyperSTL, including STL, HyperLTL, LTL and non-temporal specifications.

The modelling of electricity production and demand requires highly specific and comprehensive meteorological data. One challenge is the high temporal frequency as electricity production and demand modelling typically is done with hourly data. On the other side the European electricity market is highly connected, so that a pure country-based modelling is not expedient and at least the whole European Union (EU) area has to be considered. Additionally, the spatial resolution of the data set must be able to represent the thermal conditions, which requires high spatial resolution at least in mountainous regions. All these requirements lead to huge data amounts for historic observations and even more for climate change projections for the whole 21 st century. Thus, we have developed the aggregated European wide climate data set SECURES-Met that has a temporal resolution of one hour, covers the whole EU area and other selected European countries, has a reasonable size but considers the high spatial variability.

We present amperometric sensors based on direct electron transfer enzyme (DET) for the detection of lactate, which is an important medical parameter present in blood and interstitial dermal fluid (ISF). For measurement in blood, we present aplanar screen-printedbiosensor withcarbon working electrodes, whilefor the intendedmeasurement in ISF, we investigated platinum-metallized epoxy microneedle sensors. On both sensor types a bioink was applied, consisting of a DET enzyme mixed with poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS). As second layer, a hydrogel layer is deposited to hold the enzyme on site. Local modification of the platinum microneedle sensors was performed by non-contact spotting. The developed modification enables the detection of lactate at a potential of 0 V with response times of 500-700 s. For carbon sensors, a limit of detection of 0.12 mM lactate was determined, and two linear ranges of 0.3-5 mM and 10-50 mM were observed with sensitivities of 319 and 9.6 nA/(mm2·mM), respectively. For locally modified platinum microneedle sensors two linear ranges of 0.3-2.5 mM and 5-30.5 mM were observed with sensitivities of 322.5 and 3.7 nA/(mm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> ·mM), respectively. Given the low sensitivities in the higher concentration range, saturation for carbon sensors and locally modified platinum microneedle sensors starts at 10 mM and 5 mM lactate, respectively. Thus, both sensors allow sensitive measurements in the lower concentration range. Current densities at saturating lactate concentration are higher on freshly prepared carbon electrodes with 1.80 μA/mm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> (10 mM) compared to platinum microneedle electrodes with 0.75 μA/mm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> (10 mM) with full electrode modification. For platinum microneedle electrodes with optimized, dried local microneedle modification a current density of 0.95 μA/mm2 (5 mM) was measured. Detection of lactate in whole blood was demonstrated on carbon sensors, showing increasing currents after exercise, correlating with higher blood lactate levels, measured with a test strip reference system.

Magnesium-ion batteries (MIBs) are of considerable interest as environmentally more sustainable, cheaper, and safer alternatives to Li-ion systems. However, spontaneous electrolyte decomposition occurs due to the low standard reduction potential of Mg, leading to the deposition of layers known as native solid electrolyte interphases (n-SEIs). These layers may inhibit the charge transfer (electrons and ions) and, therefore, reduce the specific power and cycle life of MIBs. We propose scanning electrochemical microscopy (SECM) as a microelectrochemical tool to locally quantify the electronic properties of n-SEIs for MIBs. These interphases are spontaneously formed upon contact of Mg metal disks with organoaluminate, organoborate, or bis(trifluoromethanesulfonyl)imide (TFSI)-based electrolyte solutions. Our results unveil increased local electronic and global ionic insulating properties of the n-SEI formed when using TFSI-based electrolytes, whereas a low electronically protecting character is observed with the organoaluminate solution, and the organoborate solution being in between them. Moreover, ex situ morphological and chemical characterization was performed on the Mg samples to support the results obtained by the SECM measurements. Differences in the electronic and ionic conductivities of n-SEIs perfectly correlate with their chemical compositions.

Combined cyber and physical attacks on Critical Infrastructures (CIs) have disastrous consequences on economies and in social well-being. Protection and resilience of CIs under combined attacks is challenging due to their complexity, reliance on ICT systems and the inter-dependencies between different types of CIs. The PRAETORIAN framework was designed to address these challenges, by integrating components responsible for detecting both cyber and physical threats. Additionally, it forecasts how the combined attacks will evolve and their cascading effects on interdependent CIs. The PRAETORIAN framework was demonstrated based on a realistic scenario in the Zagreb airport, combining both physical and cyber attacks.