Many of today’s mechatronic systems – such as automobiles, automated factories or chem- ical plants – are a complex mixture of hard- ware components and embedded control soft- ware, showing both continuous (vehicle dynam- ics, robot motion) and discrete (software) behav- ior. The problems of estimating the internal dis- crete/continuous state and automatically devising control actions as intelligent reaction are at the heart of self-monitoring and self-control capabil- ities for such systems. In this paper, we address these problems with a new integrated approach, which combines concepts, techniques and for- malisms from AI (constraint optimization, hid- den markov model reasoning), fault diagnosis in hybrid systems (stochastic abstraction of contin- uous behavior), and hybrid systems verification (hybrid automata, reachability analysis). Prelimi- nary experiments with an industrial filling station scenario show promising results, but also indicate current limitations.     «

Many of today’s mechatronic systems – such as automobiles, automated factories or chem- ical plants – are a complex mixture of hard- ware components and embedded control soft- ware, showing both continuous (vehicle dynam- ics, robot motion) and discrete (software) behav- ior. The problems of estimating the internal dis- crete/continuous state and automatically devising control actions as intelligent reaction are at the heart of self-monitoring and self-control capabil- ities for such systems. In...     »