What's ahead for embedded software?

Once deemed too small and retro for research, embedded software has grown complex and pervasive enough to attract the attention of computer scientists. There are many research questions, but most center around one issue: how to reconcile a set of domain-specific requirements with the demands of interaction in the physical world. How do you adapt software abstractions designed merely to transform data to meet requirements like real-time constraints, concurrency, and stringent safety considerations? The answer to this question has given rise to some promising research angles, including novel ways to deal with concurrency and real time and methods for augmenting component interfaces to promote safety and adaptability.