Platform-Aware Synthesis of Embedded Control Software

There is a large semantic gap between control algorithms - making idealistic assumptions on the implementation platform (e.g., fault-tolerance architecture, memory requirement, data-rate requirement, etc.) and their actual implementation on concrete platforms. In this research area, we focus on providing controller code enforcing rich specifications while taking into account the features of the computational platforms.

Highlights of the proposed results include:

• We proposed a novel approach to design controller that provides safety guarantee on the physical component in the presence of application-level and system-level faults through the full system restart.
• We provide a tool dtControl which provides the memory-efficient, compact, understandable representation and the efficient determinization for the formal symbolic controllers obtained from the state-of-the-art toolboxes SCOTS and UPPAL. It also provides a scheme to design non-uniform quantizers (i.e., state encoders with non-uniform partitioning of state-set) for symbolic controllers. The results are also useful for constructing efficient static coders minimizing bit rate over the sensor-controller channel.