The RACE Project: An Informatics-Driven Greenfield Approach to Future E/E Architectures for Cars

Abstract

As cars are turning more and more into “computers on wheels,” the development foci for future generations of cars are shifting away from improved driving characteristics toward features and functions that are implemented in software. Classical decentralized electrical and electronic (E/E) architectures based on a large number of electronic control units (ECUs) are becoming more and more difficult to adapt to the extreme complexity that results from this trend. Moreover, the innovation speed, which will be dictated by the computer industry’s dramatically short product lifecycles, requires new architectural and software engineering approaches if the car industry wants to rise to the resulting multidimensional challenges. While classical evolutionary architectures mapped the set of functions that constitute the driving behavior into a coherent set of communicating control units, RACE (Reliable Control and Automation Environment) is an attempt to redefine the architecture of future cars from an information processing point of view. It implements a straightforward perception-control/cognition-action paradigm; it is data centric, striking a balance between central and decentralized control. It implements mechanisms for fault tolerance and features plug-and-play techniques for smooth retrofitting of functions at any point in a car’s lifetime.