Periodic Reporting for period 2 - AIDOaRt (AI-augmented automation for efficient DevOps, a model-based framework for continuous development At RunTime in cyber-physical systems)
Periodo di rendicontazione: 2022-04-01 al 2023-03-31
Modern systems in the domains of Industry 4.0 health care, autonomously driving cars, or smart grids are examples of highly communicating (embedded) systems where software enables increasingly advanced functionality. The growing complexity of these Cyber-Physical Systems (CPS) and Cyber-Physical Systems of Systems (CPSoS) poses several challenges throughout all system design, development and analysis phases, but also during their deployment, actual usage, and future maintenance. Even if more companies are deploying AI in some specific parts of their business, there has not been any general and reusable AI-augmented approach intending to support full continuous software and system engineering processes in the context of different use cases and application domains.
AIDOaRt wants to impact organizations where continuous deployment and operations management are standard operating procedures. DevOps teams may use the AIDOaRt solutions to analyze event streams (for real-time and historical data) together with the design information (e.g. in different system models) to extract meaningful insights for system continuous development improvement, to drive faster deployments and foster better collaboration, and to reduce downtime. We expect an industrial uptake of AIDOaRt technologies in the development of complex systems that scales up to real systems demand with relevance for all critical applications. The integration of AI techniques can affect the whole development process. Thus, AIDOaRt aims at providing a holistic approach for continuous systems engineering that:
(i) Provides a core model-based framework to support the CPS continuous systems engineering process by benefiting from AI-augmented automation;
(ii) Enhances the corresponding DevOps tool chain by integrating the use of AI techniques (notably Machine Learning) in multiple aspects of the system development process (e.g.to support requirements, monitoring, modeling, coding and testing activities);
(iii) Supports the collection, representation and traceability of runtime data and software models (Observe), assists in the analysis of both historical and real time data in combination with design information (Analyze), and supports the automation of tasks of the DevOps pipeline according to the previous analysis (Automate).
The AIDOaRt mission is to create a framework incorporating methods and tools for continuous software and system engineering and validation leveraging the advantages of AI techniques (notably Machine Learning) in order to provide benefits in significantly improved productivity, quality and predictability of CPSs, CPSoSs and, more generally, large and complex industrial systems.
The project will have a significant impact on the partners’ competitive advantage, growth, and internationalization. As the market grows, competition intensifies. Long-term investments in automated quality measurement and prediction as well as the utilization of artificial intelligence in test design and analysis are the focus areas of product development. Cooperation with international actors in the field is also of great benefit. We consider four different kinds of expected impact:
AIDOaRt is an enabler
1) for Large Enterprises since the average expected impact on human resources is around 5-6 new personnel hired.
2) for SMEs since it improves competitiveness in terms of quality of products and services offered, responsiveness to customers' needs as well as time to market.
3) for new markets since the expected turnover increment is evaluated between +10% up to +20% within two years after the end of the project.
4) and opportunity for academia to improve their presence, research, and activities in the project-related scientific areas.
AIDOaRt wants to impact organizations where continuous deployment and operations management are standard operating procedures. DevOps teams may use the AIDOaRt solutions to analyze event streams (for real-time and historical data) together with the design information (e.g. in different system models) to extract meaningful insights for system continuous development improvement, to drive faster deployments and foster better collaboration, and to reduce downtime. We expect an industrial uptake of AIDOaRt technologies in the development of complex systems that scales up to real systems demand with relevance for all critical applications. The integration of AI techniques can affect the whole development process. Thus, AIDOaRt aims at providing a holistic approach for continuous systems engineering that:
(i) Provides a core model-based framework to support the CPS continuous systems engineering process by benefiting from AI-augmented automation;
(ii) Enhances the corresponding DevOps tool chain by integrating the use of AI techniques (notably Machine Learning) in multiple aspects of the system development process (e.g.to support requirements, monitoring, modeling, coding and testing activities);
(iii) Supports the collection, representation and traceability of runtime data and software models (Observe), assists in the analysis of both historical and real time data in combination with design information (Analyze), and supports the automation of tasks of the DevOps pipeline according to the previous analysis (Automate).
The AIDOaRt mission is to create a framework incorporating methods and tools for continuous software and system engineering and validation leveraging the advantages of AI techniques (notably Machine Learning) in order to provide benefits in significantly improved productivity, quality and predictability of CPSs, CPSoSs and, more generally, large and complex industrial systems.
The project will have a significant impact on the partners’ competitive advantage, growth, and internationalization. As the market grows, competition intensifies. Long-term investments in automated quality measurement and prediction as well as the utilization of artificial intelligence in test design and analysis are the focus areas of product development. Cooperation with international actors in the field is also of great benefit. We consider four different kinds of expected impact:
AIDOaRt is an enabler
1) for Large Enterprises since the average expected impact on human resources is around 5-6 new personnel hired.
2) for SMEs since it improves competitiveness in terms of quality of products and services offered, responsiveness to customers' needs as well as time to market.
3) for new markets since the expected turnover increment is evaluated between +10% up to +20% within two years after the end of the project.
4) and opportunity for academia to improve their presence, research, and activities in the project-related scientific areas.
During the first part of the project, the performed work focused on different complementary aspects aiming at triggering the project’s research and development activities:
1. Collection and expression of the various (industrial) use case requirements, as well as definition of corresponding and more detailed use case scenarios; Development of use case scenarios resulting in first evaluations and results.
2. Specification of a first and second/final version of the overall AIDOaRt architecture, including a high-level specification of the three projects tool sets (Data Engineering Tool Set, Core Tool Set, AI-Augmented Tool Set);
3. In parallel, study of the scientific state-of-the-art in the main areas of interest of the project (namely Modeling / Model Driven Engineering, DevOps and Artificial Intelligence / MachineLearning);
4. In ADIOaRt the use cases and solutions are represented in an overall data model, the AIDOaRt platform / framework, defined in a Universal Modeling Language tool called MODELIO. Applying this representation has helped the project partners systematically map the complete range of Use Cases with potential Solutions. During Year 2 these mappings have been refined and followed by the first concrete applications of solutions to matching Use Cases.
5. In the context of previous actions, various research experimentations (resulting in some published scientific papers) and technical developments (resulting in technical solutions improvements). More practical works and numerous partners collaborations fostered by three project internal hackathons.
1. Collection and expression of the various (industrial) use case requirements, as well as definition of corresponding and more detailed use case scenarios; Development of use case scenarios resulting in first evaluations and results.
2. Specification of a first and second/final version of the overall AIDOaRt architecture, including a high-level specification of the three projects tool sets (Data Engineering Tool Set, Core Tool Set, AI-Augmented Tool Set);
3. In parallel, study of the scientific state-of-the-art in the main areas of interest of the project (namely Modeling / Model Driven Engineering, DevOps and Artificial Intelligence / MachineLearning);
4. In ADIOaRt the use cases and solutions are represented in an overall data model, the AIDOaRt platform / framework, defined in a Universal Modeling Language tool called MODELIO. Applying this representation has helped the project partners systematically map the complete range of Use Cases with potential Solutions. During Year 2 these mappings have been refined and followed by the first concrete applications of solutions to matching Use Cases.
5. In the context of previous actions, various research experimentations (resulting in some published scientific papers) and technical developments (resulting in technical solutions improvements). More practical works and numerous partners collaborations fostered by three project internal hackathons.
The progress beyond the state-of-the-art affects data management and storage, Model-Driven Engineering for continuous software and system engineering, and AI/ML for continuous software and system engineering. The AIDOaRt framework will support inherent heterogeneity and the fast-paced evolution of a continuous collection of engineering data, providing both process- and product-related information, tailoring its data collection and management services to the analysis needs of other technical areas. It also addresses the main challenge of introducing an MDE process for DevOps, augmented with AI/ML mechanisms.
The foreseen socio-economic impact AIDOaRt is expected to contribute to two of the most challenging sectors i.e. automotive and transportation (long term impact 17%) focussing on productivity gain obtained by higher automation of the software engineering process, which is a shared goal of AIOps (AI applied to DevOps) and MDE.
Such impact, in conjunction as well with the expected results, is also presented by exploitation and communication work.
The project website as the main presence of project outlet has achieved over 4770 unique per year
The followers of online social media community (LinkedIn and Twitter) is over 300
The impact in the scientific community includes 4 research papers in journals 8 papers in conference proceedings
The exploitation and standardization plans and actions performed during the reporting period to analyze the potential AIDOaRt partners' exploitation assets and the agreed individual and joint exploitation strategies as they have been included in our Project Consortium Agreement were delivered. Also, along this period, partners of AIDOaRt have been working in the standardization efforts for MARTE and SysML, two of the modeling standards mentioned in the proposal.
The foreseen socio-economic impact AIDOaRt is expected to contribute to two of the most challenging sectors i.e. automotive and transportation (long term impact 17%) focussing on productivity gain obtained by higher automation of the software engineering process, which is a shared goal of AIOps (AI applied to DevOps) and MDE.
Such impact, in conjunction as well with the expected results, is also presented by exploitation and communication work.
The project website as the main presence of project outlet has achieved over 4770 unique per year
The followers of online social media community (LinkedIn and Twitter) is over 300
The impact in the scientific community includes 4 research papers in journals 8 papers in conference proceedings
The exploitation and standardization plans and actions performed during the reporting period to analyze the potential AIDOaRt partners' exploitation assets and the agreed individual and joint exploitation strategies as they have been included in our Project Consortium Agreement were delivered. Also, along this period, partners of AIDOaRt have been working in the standardization efforts for MARTE and SysML, two of the modeling standards mentioned in the proposal.