14. – 19. Mai 2017, Dagstuhl Seminar 17201
Formal Synthesis of Cyber-Physical Systems
Matthias Rungger (TU München, DE)
Auskunft zu diesem Dagstuhl Seminar erteilen
Susanne Bach-Bernhard zu administrativen Fragen
Marc Herbstritt zu wissenschaftlichen Fragen
(Zum Einloggen bitte Seminarnummer und Zugangscode verwenden)
Cyber-Physical Systems (CPS) are complex systems resulting from intricate interaction of discrete computational devices with the continuous physical plants. Within CPS, embedded control software plays a significant role by monitoring and adjusting several physical variables, for example, temperature, velocity, pressure, density, and so on, through feedback loops where physical processes interact with computational devices.
With the growing trend in computational devices, embedded control software is becoming more and more ubiquitous in many safety-critical applications including automotive, aerospace, transportation systems, critical infrastructure, energy, robotics, healthcare, etc. Unfortunately, the design of embedded control software nowadays is still based on ad-hoc solutions resulting in brittle and error-prone software, and very high verification and validation costs.
In order to detect and eliminate design flaws and inevitable software bugs, a large portion of the design budget is consumed with validation and verification efforts, which are often lengthy. On the other hand, by changing the emphasis from verification to synthesis, it is possible to synthesize correct-by-design embedded control software for CPS while providing formal guarantees of correctness and preventing the need for costly post facto verifications.
This seminar focuses on automatic, correct-by-construction, and scalable synthesis of controllers for interacting discrete and continuous systems, namely, hybrid systems. It will bring together an inter-disciplinary team of computer scientists, control theorists, and researchers from industry to identify a unified methodology for controller synthesis of CPS and discuss potential issues especially raised in applying those techniques to industrial platforms, which require the tight collaboration between these communities. Towards achieving this goal, a merging of techniques between (continuous) control theory and (discrete) computer science is discussed in the following key areas:
1. Abstraction-based synthesis;
2. SMT-based synthesis;
3. Reactive synthesis;
4. Learning-based synthesis.
The research issues relevant to the synthesis of cyber-physical systems will be discussed during the seminar, including:
1. Scalability of the synthesis techniques for hybrid systems;
2. Platform-aware synthesis of hybrid systems;
3. Unifying concepts such as robust synthesis for discrete and continuous systems;
4. Distributed synthesis of hybrid systems.
This seminar will benefit the control as well as computer science communities by bridging the gap between the complementary concepts. It will be a unique opportunity for an active and productive interaction between these communities.
Creative Commons BY 3.0 DE
Calin A. Belta and Rupak Majumdar and Majid Zamani
- Artificial Intelligence / Robotics
- Modelling / Simulation
- Semantics / Formal Methods
- Formal Synthesis
- Reactive Synthesis
- Correct-By-Construction Synthesis
- Cyber-Physical Systems
- Hybrid Systems
- Control Systems