There is already effective use of computers in vehicle control systems such as electronic ignition, engine management and anti-skid brakes, and even more complex applications are being investigated. Ideas being seriously considered in DRIVE include the "intelligent" co-pilot, collision avoidance, lane control and road access control. However, given the complexity of these systems, one is forced to question our reliance on computers in systems whose failure could result in an accident. In other words, can we trust the computer? Clearly, such a fundamental question requires at least a partial solution before the more advanced, safety-critical computer-based systems can be accepted for general usage. We may be faced with a technology which allows us to build systems of great complexity and potential benefit, but whose likelihood of failure we are unable to predict. This problem is not unique to road transport. In this paper, the problem is examined in more detail, and in the context of RTI (Road Transport Informatics) systems. The paper is a product of DRIVE V 1051, which comprises TÜV Rheinland (Germany), TNO (Netherlands), Leeds University (UK) and Program Validation Limited (UK).
Abstract