Autonomous systems are entities which are able to work properly without (or very limited) human intervention, especially when facing unknown environments and/or unexpected events (e.g. failures). They are already a major topic for AI research, mostly in robotics and space applications.
The notion of "self-*" (pronounced self-star) has been introduced in software engineering, referring to concepts and capabilities which enable an entity to reflect and act on its own processes in order to increase its autonomy. Among the self-* properties, we can distinguish properties like self-awareness, self-adaptiveness, self-correction, self-healability, self-organization, self-repair, etc. The main applications are autonomic information systems, overlay networks, web services, etc.
Self-star and autonomous systems hence designate similar concerns, which can be met in a lot of other application fields (avionic systems, automotive systems, automatic production, etc). Designing such systems requires high-level reasoning techniques, which are usually addressed within the AI community, such as dynamic online planning (to adapt one's behaviour to a new situation), reactive replanning (to simply repair the current behaviour of the system), online diagnosis (to infer the actual state of the system from what can be observed) or any other techniques closely related to the specific self-* properties that are requested by the system.
Moreover, such techniques must be closely integrated, to get smooth and safe continuous working conditions. Monitoring self-* systems amounts to an online decision policy which must be continuously run to decide what to do next: wait for more observations, compute a diagnosis, plan for some next stage, plan for repair, plan for active diagnosis, etc. This closer integration requires important extensions to the classical diagnosis reasoning techniques (diagnosis objectives? diagnosability issues), planning reasoning techniques (continuous robust planning under uncertainty, study of the needs for replanning and recovery planning), and decision processes analysis (considering different types of actions, temporal constraints, etc).
The purpose of this workshop is to gather researchers from different AI fields in order to fill the gaps and exchange ideas about which reasoning techniques are needed for self-* and autonomous systems. Possible topics include but are not limited to:
Workshop MotivationThe purpose of this workshop is to gather researchers from different AI fields in order to fill the gaps and exchange ideas about which reasoning techniques are needed for self-* and autonomous systems (diagnosis, planning, reasoning about uncertainty).
This workshop will be organised as a set of thematic sessions (with 2-4 papers per session). The papers will be disseminated among the participants prior to the workshop, and the organizers will appoint for each paper an opponent among the other participants. The paper will be allocated around 20 minutes for the presentation, and the opponent will be given around 5 minutes to give her/his own comments on the work, then 5 minutes will be left for a couple of technical / clarification questions. The opponent will be someone having interest in the field but not necessarily working in the same area (e.g. a researcher working in dynamic planning will have to comment on a paper dealing with diagnosing plans). Then each session will end with a common 30 minutes period for questions and debate around the theme of the session.
As a final session, we also propose that the organizing chairs summarize the contributions of the workshop in order to emphasize on the fundamental ideas about how to use AI in self-star systems. This small talk will be followed by one hour discussion and debate in order to provide a synthetic conclusion to the workshop, and possibly pave the way for specific collaborations and projects in that field among some of the participants.
Workshop AttendanceThe workshop is open to all members of the AI community but the number of participants is strictly limited to 75 participants.
* denotes the workshop co-chairs.