Same domain
See Also
- Optimisation
- Probability in modelling, algorithms, and verification
- Uncertainty
- Interactions, visualisation and virtual reality
- Decision and control systems
- Sensing systems
- Computing systems
- Modelling and simulation
- Software
Complements
Hybrid systems
This is about designing and validating systems that combine continuous-time parts (associated with, e.g., the physical world) and event-driven parts (associated, e.g., with mbedded computers).
Combustion engines, for instance, are hybrid systems per se, as the nature of the continuous-time physical processes taking place in the cylinders depends on the stroke (intake, compression, combustion, exhaust), which itself depends on conditions on the engine state.
Most often, however, hybrid systems involve computers that may be embedded in the part of the physical world with which they interact and may also be connected to it and together through networks.
The number of such hybrid systems explodes as computers interconnect at a staggering rate with physical systems that they contribute to make more adaptable and more efficient.
Design is necessarily multidisciplinary and multi-trade. It relies on a variety of mathematical models (PDEs, ODEs, discrete-events, hybrid), where control theory, computer science, signal processing and statistics play fundamental roles.
The main challenge is to develop and disseminate methodologies to increase robustness and performance while reducing time and cost of design, development, validation, production and maintenance.
Since many hybrid systems are safety critical (transportation or life-care systems, for instance), reliability is a major issue. Cooperation between Digiteo groups should bear fruit in many directions.
Here is a non-limitative list of examples:
- Foundational multi-disciplinary research (Control, Computer and Communication) and proofs of concept addressing the whole chain from observers, sensing, monitoring and actuation to adaptive and cooperative monitoring and control and decision making need to be developed for hybrid, possibly networked, systems. Parameter uncertainty, delays, disturbances, limited communications bandwidth, actuation constraints and node availability must be addressed.
- Global approaches that address the design of the system as a whole, with its many physical, functional and logical aspects, taking into account the constraints inherent to embedded systems, should be encouraged.
- New possibilities offered by computer sciences may impact the very nature of control algorithms. The safe implementation of control laws requires dedicated operating systems, dynamical reconfiguration of architectures and scalable algorithms for the control of evolvable, distributed and adaptable systems.
- Formal tools that are being developed in computer science for analysis and proof of software bring new possibilities to deal with models of physical systems where time plays a critical role. To extend static analysis to hybrid systems, for instance, it is necessary to assess the range of the values taken by variables of uncertain dynamical models of physical systems, e.g., via interval analysis and guaranteed integration.
- Besides the current application domains (including transportation, energy management, environment monitoring, factory automation, personal communication, process industry), new application domains should be explored.
