Designing intelligent control systems for safety critical applications

This paper presents a design methodology for behaviour-based intelligent control systems, which we argue is suited to safety critical applications. The methodology has a formal basis, and uses dynamical systems theory to prove system safety properties that are expressed in terms of Lyapunov stability. We propose a new computational model for implementation of these systems, which offers a reduction in complexity due to its non-symbolic structure. Reduced complexity is attractive because it allows improved depth of failure analysis, and potentially increased system reliability.