Team: Westermo + RISE
Challenge: Identifying bug-inducing code changes
- In the DevOps process at Westermo, developers commit code changes regularly, and we do nightly testing of the software running in our networked embedded systems. The nightly testing is run by the Fawlty test framework -- built in-house. Test results are stored in a test results database.
- The goal of this challenge is to identify links between in-data for nightly testing and out-data from it, when considering code changes in the two repositories (test framework, and software) as in-data, and test results as out-data. The end goal of this work is tobe presented with a list of likely bug-inducing changes when one comes to work in the morning and looks at a number of failing tests.
Team: Westermo + COPADO
Challenge: Test case dependencies
- In the nightly testing at Westermo, test results data is stored in a test results database. In this challenge we will continue exploring test case dependencies from this data. This data has been released at github: https://github.com/westermo/test-results-dataset External link.
In previous AIDOaRt hackathons, we have seen that there seem to be dependencies, correlations, and/or inference/association rules that can be identified between non-passing tests. One could imagine a test case A that sometimes fails to clean up the system if it crashes, so that test case B always fails if A also fails.
In this hackathon, we wish to explore this further, and with open data:
- What is the best way to identify links between failing tests?
- How can these links be understood, when looking at them over time? Are some dependencies reinforced or removed?
- What is the best way to visualize links such that a Westermo colleague can look at test results and understand that there might be links?
- Can the links be identified under simulated resource constraints (where one can only run a third of the test cases each night?)
Team: VCE, IMTA, JKU, UNIVAQ, MDU
Challenge: Modeling recommendations and process mining for system architecture descriptions
Continuation of the previous hackathon to develop a solution architecture for the development of VCE system architecture descriptions with the added capabilities of process mining and modeling recommendations.
- Goal with the hackathon is the realization of concrete recommendations for the system architectures in use, currently the solution is not working towards the concrete artifacts of VCE. A positive outcome of this activity would be the a functional prototype that can provide meaningful recommendations.
Team: VCE, DT, MDU
Challenge: Continuous delivery of FMU
- This challenge aims to include DevOps into the current workflow through the context of simulation. The challenge is particularly focused on the FMI 2.0 standard and the use of corresponding FMU units developed through modelling in tools such as Simulink
- The goal is the integration of a CD pipeline for FMU corresponding to different variants of one or more components to enable simulation for early analysis. It is foreseen that the use of FMI could be enable for co-simulation which could enable analysis of machines at an earlier stage with a library of FMUs.
Team: VCE, MDU
Challenge: Architecture optimization
- This challenge is a continuation of a previous challenge from the 2nd hackathon. Generally, the idea is to implement an automation system that will assist in architecture modeling of a thermal management system at Volvo CE. In this regard it is also of interest to understand the best way to model such a system within the scope of Volvo CE.
- In this challenge, we are interested in automate the creation of the layout as shown in the Figure above. There are numerous variants of each component (component of the TMS, battery, dc/dc, etc,.). Each machine has their own variants of components such as the battery, dc/dc converter. The goal is to optimize the layout based on coolant flow and pressure properties of each component in the coolant loop.
Team: AVL, TUG
Challenge: Learning-based fuzzing of AGL
- Automotive Grade Linux (AGL) is an open-source operating system for automotive applications. The AGL project is trying to create a standard for in-vehicle software solutions by bringing together car manufacturers, suppliers, and software companies. Despite the fact that AGL is still very much in development, the first distributions are already available. To be part of this standardization process, it is critical to ensure the reliability and security of such systems from the very beginning.
- The goal of this challenge is to test the different communication interfaces of an AGL demonstrator. Assuming a real scenario, we consider the AGL demonstrator as a black box where we can only execute inputs and observe the corresponding outputs. To address the challenge of an unknown state space, we aim to automatically build a behavioral model of the communication interface using model learning. This learned model is then the basis for creating a stateful black-box fuzzer.
Challenge: Model Based Testing of an Autonomous Driving System
- Advanced Driver Assistance Systems (ADAS) are systems that are capable of handling specific driving tasks such as highway driving. Those tasks are complex, and mistakes can incur high costs and in the worst case human casualties. Consequently, a lot of effort in the development of ADAS has to be spent on ensuring the safety of the system. An important part of this is the definition of reference scenarios in which the system should behave safely, which can for example be done using the OpenSCENARIO file format that combines static and dynamic environment information for simulation tailored to the automotive domain. It allows the specification of complex, parameterized and interactive manoeuvres involving several traffic participants.
- The goal of this challenge is to explore the viability of time-based approaches in the context of ADAS testing. This involves:- Definition of KPIs for individual simulation time steps.
- Apply behaviour-based methods using previously defined KPIs to search / identify critical regions for a given action-based scenario.
- Analyzing situations in which the current method and the corresponding KPI definitions struggle.
Team: Alstom + MDU + RISE + SOFTEAM
Challenge: Requirements analysis, processing and response
- Alstom’s engineers need to evaluate and answer customer requirements in a short time. The analysis of the requirements is extensive and time-consuming. The engineer has to allocate them to different teams, assign the person responsible for them, and analyze whether Alstom can comply with the given requirement. Each project contains several thousands of both hardware and system requirements. These requirements are stored in DOORS Rational Software, and they are going through several phases of requirements analysis.
- Alstom’s goal is to minimize time spent on analyzing the requirements and maximize the number of correctly evaluated and answered requirements.
- To date the following SW tools / SW modules are under development as a result of a previous hackathon:
- Requirement Identifier
- Ambiguity Checker
- Requirements Semantic Search
- Team Allocator
Team: BT/ALSTOM + AVL+ABO
Challenge: Automatic parametrization of PS controller
- Parametrizing control models of the propulsion system is done manually during the physical system testing
- Alstom’s control/test engineers need to automatically parametrize control models of the propulsion system
- Automate the parametrization process using AI/ML models
- Minimize time and effort spent on parametrizing control models
Team: PRO + ACO + ITI + UOC
Challenge: Monitoring and modelling for a smart platform
- The Smart Port Monitoring Platform (SPMP) involves the deployment of a large infrastructure of interconnected elements that need to be continuously monitored so that any failure that occurs can be detected and corrected as soon as possible so that the platform does not interrupt its workflow (or only minimally). Therefore, it is necessary to implement systems for anomaly detection and it is also vital to model the infrastructure in advance in order to have the necessary resources available.
- Relevant sensing and monitoring requirements for the detection of anomalies. Requirements on positioning solution.
- Define the base model of the SPMP and possible scenarios.
- Detecting anomaly patterns
Team: ABINSULA, INTECS, UNI of SASSARI
Challenge: Adoption of AI and ML techniques for image processing in the automotive context
- Modern cars are connected systems that acquire inputs from the environment and are expected to autonomously External link. react External link. according to external stimuli External link. and internal needs; thus, they can be considered as Cyber-Physical Systems (CPS). The Abinsula case study presents a virtual rear-view mirror scenario in which multiple cooperative cameras are used to capture the context outside the vehicle, by means of AI-based technology. This is something allowed only in concept cars and small productions that do not apply the same regulations of large productions.
- One of the main challenges we address in this use case, is the replacement of car mirrors with cameras. Therefore, we want to study the adoption of AI and ML techniques for image processing to detect and to signal possible hazards, considering their possible implementation on FPGA boards.
Team: AVL + JKU + UCAN
Challenge: Optimization of Development Processes
- Traceability among data artifacts is imperative to manage the development of complex technical systems, particularly modern vehicles. In the next few years, the implementation of cross-discipline traceability in industry data management solutions will lead to massively linked and highly structured data. The issue is: This structured data is currently in the very process of being created and not yet fully available. But the mining solutions need to be developed now to be available on the market once structured data is available. Thus, the key question is: How to artificially create realistic structured data?
- We need an algorithm that can emulate vehicle development projects (VDP) and produce realistic structured data.We focus on the part of product development that deals with iterative product optimization. The algorithm’s key capability must be to iteratively optimize parameters (e.g., aerodynamic drag) to eventually make all product CVs meet the TCVs (Target Characteristic Values, e.g., energy consumption).
- Two goals are defined for this hackathon:
- Capture a small fragment of a generic automotive development process as an SPEM-based process model in UCAN’s process modeling and knowledge base solution component
- Capture the same small process fragment as a set of model parameters and model transformations by means of an Ecore-based process model in JKU’s MOMoT solution component.
Challenge: Runtime Verification
- Integration of AND’s tools into HIB’s TAMUS development environment was started in the first hackathon but never completed (signing an NDA is no easy task). But now that is solved so we jump into full automation for CD and CI
- Analyze current state of practice for TAMUS development wrt CD and CI.
- Identify tools in AND’s portfolio to help solve low hanging fruit for now.
Challenge: Transforming Formal Specifications to Application Code for Railways Cyber Physical Systems
- Challenges in transforming formal specifications to application code for railways CPS in B language.For many components, refinement is a simple routine that could probably be written directly from abstraction. In the past, there have been attempts to automatically refine specifications, for example using patterns, but the technology was not flexible enough to deal with the variability of human written specifications. Now, we believe that AI can complete the task.B language works in two falvours: specification and code.
- An effective and accurate AI solution for transforming formal specifications to application code for cyberphysical systems.
- Solution prototypes developed during the hackathon that can be used for future applications.
- A community of participants and experts who can continue working together to improve AI solutions for cyberphysical systems.