Computation Aware Control Systems – COMPACS

An important application domain of embedded technology is the field of control systems where embedded software is executed in order to control a physical plant. Traditionally, the design of embedded control systems often assumes a separation of concerns between computation and control. An integrated approach where the constraints due to limited or shared computational resources are taken into account in the synthesis of control strategies would enable the development of high quality embedded controllers with guarantees of safety, stability and performance, while optimizing the usage of the available computational resources. This line of research, developed in the project COMPACS, is relevant to the field of cyber-physical systems, which pursues a better integration of computational devices with physical processes.

More precisely, the COMPACS project aims at pursuing fundamental research towards the development of sound model-based approaches for the synthesis of controllers that are aware of the computational resources. Computation aware control strategies should not only be able to decide which control actions need to be taken, but also when the control tasks should be scheduled based on the availability of computing units and the state of the physical plant. These strategies are of particular interest when the dynamics of the physical plant have multiple time-scales and the computational resource should be allocated more often to control the faster dynamics than the slower one. COMPACS will tackle this problem in the modeling framework of hybrid systems, which enables an appropriate description of the heterogeneous nature of computation aware control. The project brings together hybrid systems experts, from different disciplines (LJK: applied mathematics, CRAN: control theory, VERIMAG: computer science), and is organized in two main technical work-packages. The first work-package investigates fundamental problems in hybrid systems, which will have direct applications in the development of design techniques for computation aware control systems; these techniques form the subject of the second work-package:

1. Fundamental problems in hybrid systems will be tackled: reset control systems, multiple time scale systems, discrete/continuous controller co-design for hybrid systems. These problems have all been selected because of their relevance to computation aware control systems design but are also important from the point of view of hybrid systems theory. They will be investigated using approaches based on Lyapunov techniques, reachability analysis, or a combination of these. The related numerical aspects will also be considered.

2. Computation aware control design is the main objective of the project. We will consider the problem of deciding when control tasks need to be executed within the event-based control paradigm, which enables a reduced usage of the computational resources when compared to classical time-triggered implementation. We will also consider the problem of scheduling and coordination of control tasks when the computational resource is shared or distributed. Using a description of these problems in the hybrid system framework, we will address them using the techniques developed in the first-work package.