REGISTER

Neil Robinson

FIEAust, PhD, BSc(Hons)

RGB Assurance

I started in signalling more than 30 years ago at British Rail, where I learnt how to design interlockings, initially in relay circuits, and then by programming Solid State Interlockings. This work sparked my interest in safety critical systems. My career since then has taken me into universities, where I worked on formal methods for safety-critical systems, and then back into industry. At the University of Queensland, in Australia, I was lucky enough to work on a project with Queensland Rail to produce a toolset to design and formally verify Control Tables. When we achieved a formal verification of the Control Tables for Roma Street station, I thought that the problem of ensuring safe signalling designs was basically solved. There have been other similar projects across the world that have developed tools that can formally verify the most complex signalling designs. However, 30 years on from when I started in signalling, at least on the signalling projects that I have been involved in, very little has changed. The safety of signalling designs is still largely reliant on a small group of experts that check the signalling logic and perform principles testing. The same kinds of errors that I was familiar with 30 years ago are still being made and being found by those experts. Formal methods have been used on some projects across the world, but their use is still not common. Structured (not formal) verification and validation methods are relied upon in many if not most projects in Europe, but in Australia at least, we rely on traditional methods, including principles testing. In this paper I will discuss some of the innovations that have been made in verification and validation of signalling designs over the last few decades, discuss their potential benefits, and discuss the challenges to their deployment on real projects.