We are happy to announce the 3rd international DREAMS Workshop which is being organized in the context of the ICON project “Layers of Protection Architecture for Autonomous Systems” (LOPAAS), a collaboration between Fraunhofer and the University of York / Assuring Autonomy International Programme.
With the DREAMS Workshop, we want to bring together different groups to create synergies and establish a shared understanding of the challenges and solutions related to “dynamic assurance” and “dynamic risk management”. We welcome you to join the workshop and help in shaping a reference framework for assuring autonomy.
The DREAMS Workshop is co-located with the 18th European Dependable Computing Conference (EDCC) in Zaragoza, Spain from 12-15 September 2022.
Dr. Rasmus Adler and Philipp Schleiß from Fraunhofer
Dr. Richard Hawkins from the University of York / Assuring Autonomy International Programme
Autonomous systems have enormous potential and are bound to be a major driver in future economic and societal transformations. In contexts where safety, or other critical properties, need to be guaranteed, it is, however, presently not possible to exploit the full potential of autonomous systems. Unknowns and uncertainties are induced due to the high complexity of autonomous behaviors, the utilized technology, and the volatile and highly complex system contexts. These characteristics render the base assumptions of established assurance methodologies (and standards) void; hence new approaches need to be investigated.
One general approach for making autonomous systems dependable is to make them aware of risks and empower them to assess and control those risks. Implementing such a Dynamic Risk Management (DRM) approach comes with many challenges concerning the necessary self-awareness and context awereness. On the one hand, powerful and thus complex self-awareness and context awareness are necessary to minimize risks, resolve conflicting objectives, and make acceptable trade-off decisions. On the other hand, the complexity of DRM is in conflict with assurance and high confidence in adequate risk management. DRM has the potential to not only outright enable certain types of systems or applications, but also to significantly increase the performance of already existing ones. This is due to the fact that by resolving unknowns and dealing with uncertainties at runtime it will be possible to get rid of worst-case assumptions, which are typically detrimental to a system's performance properties.
The DREAMS Workshop intends to explore concepts, techniques and technology for realizing DRM.
It invites experts, researchers, and practitioners for presentations and in-depth discussions about the current status of DRM in practice, its relevance for specific use cases, its relation to existing assurance frameworks for autonomous systems, and standardization activities. DREAMS aims at bringing together communities from diverse disciplines, such as safety engineering, runtime adaptation, predictive modeling, and control theory, and from different application domains such as automotive, healthcare, manufacturing, agriculture and critical infrastructures.
Topics of interest include but are not limited to
The DREAMS workshop is co-located with the 18th European Dependable Computing Conference (EDCC) in Zaragoza, Spain, from 12-15 September 2022.
6 June 2022
|25 June 2022|
3 July 2022
All submissions will be peer-reviewed by at least three members of the program committee.
They will be evaluated based on originality, contribution to the field, technical and presentation quality, and relevance to the workshop.
Please consider the following page limits:
All submissions should describe, in English, original work that has not been published or submitted for publication elsewhere.
Papers will be evaluated based on originality, contribution to the field, technical and presentation quality, and relevance to the workshop.
PDF versions of the submissions should be submitted electronically via EasyChair.
Please note that the review process is double-blind, so papers submitted for consideration should not include the names of the authors and their affiliations.
We are in the process of negotiating the best option for the workshop proceedings. This information will be available soon at the workshop website. We request submissions to be prepared in the standard Springer LNCS conference format: Authors should consult Springer's authors' guidelines and use their proceedings templates, either for LaTeX or for Word, for the preparation of their papers.
Corresponding authors of accepted papers, acting on behalf of all of the authors of that paper, should expect that they will have to complete and sign a Consent-to-Publish form.
Videos and recordings of the talks at the DREAMS Workshop 2021
In this keynote, the research field of dynamic risk management is structured and viewed in relation to some topics of machine ethics and the ethics of risk. In this context, the Responsive Sensitive Safety Model (RSS) concept is linked to the Situation-Aware Dynamic Risk Assessment (SINADRA) concept.
The talk is about the architectural building blocks required for performing situation-aware dynamic risk assessment (SINADRA).
If you want to know more about our research on safety topics at Fraunhofer IESE, also check our webpage: Dependable AI
This talk is about handling uncertainties of data-driven models with respect to safety constraints such as RSS.
Other presentations of accepted workshop papers on the topics of “Service-Oriented Reconfiguration in Systems of Systems Assured by Dynamic Modular Safety Cases”, “Behavior Prediction of Cyber-Physical Systems for Dynamic Risk Assessment”, “Autonomic service operation for cloud applications: Safe actuation and risk management”, and “Evaluation of Human-in-the-Loop Learning based Autonomous Systems” were not recorded, so we have to refer to the proceedings.
The presentations and discussions about dynamic risk management for autonomous systems were recorded at an event organized in the context of “The Autonomous”.
Videos and recordings of the talks at the DREAMS Workshop 2020
This keynote talk given by Philip Koopman puts dynamic risk management (DRM) into the big picture of safety assurance for self-driving vehicles. It focuses on the safety case and on monitoring its validity by means of safety performance indicators. DRM relates to the “runtime safety monitor” on slide 14 (minute 27).
This invited talk by Michael Woon discusses the realization of dynamic risk management (DRM) for autonomous vehicles.
It is closely related to our research:
This keynote talk by Detlev Richter from TüV Süd discusses dynamic risk management in the context of smart manufacturing. It highlights the need for dynamic risk management and provides a clear vision for its realization.
The working group for AI in smart manufacturing (of which Detlev Richter and Dr. Rasmus Adler are members) agreed on the need for DRM and a related recommended action was formulated in the upcoming German standardization roadmap for AI.
The safety- administration shells (green boxes at minute 16:30) relate to
Digital Dependability Identities and their instantiation for functions/components/systems in the production domain, like in this success story with Sick AG. This can enable automated risk reasoning in an IT backbone (see
https://www.basys40.de/ and https://www.fab-os.org/).
This invited talk from Pai Ganesh provides an overview of dynamic assurance concepts, and an example of how these concepts can be implemented in aviation. It also refers to several research papers for further reading.
This paper presentation by Elham Mirzaei focuses on dynamic safety cases. The approach is strongly related to our research basaed on Conditional Safety Certificates (ConSerts) and Digital Dependability Identities.
Dynamic safety cases enable runtime reasoning about the safety of systems/components that are dynamically composed at runtime. It is also the basis for making a composition aware of the current risk as it enables safe shared perception.
This paper presentation by Faiz Ui Muram presents a simulation-based approach for identifying hazards during verification and validation to gain higher confidence in production-side safety. The simulation is done at design-time (during V&V) but the fundamental idea is closely related to runtime-assurance. Furthermore, the paper contributes to the topic of continuous (safety) engineering.