Feedback control is pervasive in our lives, as it often lays the
foundation of technology we daily interface with. Control systems are at
the core of the energy distribution infrastructure, govern the behavior
of vehicles, and are embedded into household appliances like washing
machines. Controllers usually use sensors, that provide information
about the physical environment, to calculate the values that actuators
should assume to fulfill specific requirements. For example, adaptive
cruise control systems use the throttle to automatically adjust the
vehicle's speed, while maintaining a safe distance from vehicles ahead.
The actuator values are calculated using hardware and software. Hence,
the computation of the new control signals is subject to accidental
faults, systematic issues, and software bugs. However, in many cases,
these computational problems can safely be ignored. This talk will
introduce the problem of analyzing the behavior of control software
subject to computational problems, and verifying when no corrective
action is needed, i.e., when the control software is able to fulfill the
system requirements despite the presence of computational problems.