Context processing for automotive human-machine interfaces

In this work, we focus on the improvement of human-machine interaction inside the automobile by reducing the complexity of involved context management. A contribution in the form of a context processing engine named probabilistic application layer (PAL) is provided, which addresses following issues: Guarantees for time bounds in performing safety-critical exact inference, standardized application interface towards application space, dynamic changes on the structure of the belief networks by adding or removing nodes while retaining time guarantees. By addressing these issues, we take the context processing away from the application developer and ease the development of the situation-aware human-machine interaction. Additionally, we present a prototypical implementation of PAL, implemented natively on an embedded platform and connected to the human-machine interface of our project vehicle prototype.