Software architecture for adaptive collision avoidance systems

Emergent Collision Avoidance Systems (CAS's) are beginning to assist drivers by performing specific tasks and extending the limits of driver's perception. As CAS's evolve from simple systems handling discrete tasks to complex systems managing interrelated driving tasks, the risk of failure from hidden causes greatly increases. The successful implementation of such a complex system depends upon a robust software architecture. Most of the difficulty in implementing system arises from interconnections between the components. The CAS architecture presented in this paper focuses on these interconnections to mitigate this problem. Moreover, by constructing the CAS architecture through the composition of existing architectural styles, the resulting system will exhibit predictable qualities. Some of the qualities represent limitations that translate into constraints on the system. Others are beneficial aspects that satisfy stakeholder requirements.