Human-Automation Interactions in Rail Network Control

Named Investigators: Dr Sarah Sharples (University of Nottingham); Prof John Preston (University of Southampton)
Researchers: Laura Millen, Nora Balfe [PhD student sponsored by Network Rail and EPSRC] (University of Nottingham); John Armstrong (University of Southampton)
Industry Collabotarion and Mentors: Network Rail; Arup

Background:
Signallers and controllers are crucial for the efficient, reliable and safe running of railways. Understanding the skills and expertise of these key workers is particularly important in the case of modern technical support systems, such as ARS, ERTMS and the prototype tools developed by Project B1, which may also require new organisational systems.
The Human Factors approach address issues arising from a belief that automation will solve operational problems and that better technical systems will improve performance. 'Ironies of automation' have been found in many industries. Automation often fails to provide the expected benefits in functionality and reliability compared to human-centred systems. Automation can fail and then recovers less well than do people. The people designers believe to be unreliable in fact keep the total system running, despite technical failures. The very skills that people can bring to a hybrid (automation plus people) system are difficult to develop and maintain if the design philosophy is for them to be monitors only. Failure of automatic systems can then be catastrophic for performance continuity. There is therefore an ongoing need for the involvement of people in complex operating systems such as rail network control.

Objectives :

To explore and define the technical and organisational systems which best allow a hybrid control option whereby safe, efficient and reliable network control is achieved through well-designed collaborative partnership between human operators and various types and level of automated system.
To identify decision and communications support that make best use of staff skills to interpret, prioritise, intervene and optimise.

Summary of methods:
The project will draw from and expand on current work with Network Rail to identify the skills, experience and needs of signalling and control (see, for example, Pickup et al., 2005) and examine these in the light of requirements of ERTMS and other systems, such as the prototypes developed by Project B1. A variety of methods will be used. Ethnographic field analysis and cognitive work analysis will be used in the early stages. They will be followed by expert panels and cognitive walkthrough to examine the roles of key workers with various levels of automation, and particularly to generate functional analyses, scenarios and roles for a hybrid system using human skills and high quality technical support. Identification and development of existing and possible future forms of technical support will be based on the work on train control carried out in B1. The cognitive work analyses, operation sequence diagrams and human reliability predictions produced will then be used to assess the effectiveness of the different hybrid system options in terms of total system reliability and performance.

Anticipated Impacts and Benefits:
The need for improved train control has been identified by Network Rail, DfT Rail and TOCs. This project will bring together the relevant human factors and train control issues, to reduce delays to trains and their users and improve the efficiency with which the railway network is used.

Work done:
UNott: This work has primarily been focussed on development of methodology and understanding of the impact of automation within rail control. The ironies of automation (Bainbridge, 1983) have been revisited and the key issues that are relevant for the modern railway control environment identified. These will be used as the basis for the analysis of ethnographic data collected in rail control environments during 2008-2009. A study has been done to assess the reliability of the observation methods and signaller work description categories to be applied in the data collection phase and this has shown an acceptable level of inter-rater reliability and validated the categories of work description to be applied. In addition, work done in the NR sponsored project has examined the relationship between different signaller activities within ARS IECC environments.

Work is now underway to apply the observation techniques to a set of signal environments and compare the data obtained with the operational demand evaluation checklist (ODEC) and Integrated Workload scale (IWS) to gain a further understanding of the relationship between different types of automation use, signalling environment and perceived levels of workload.

UoS: Work has focussed on synthesising the results of the preceding project B1 (Preston et al., 2007) to determine the possible benefits of automation. Professor Preston contributed to a workshop on Network Management Research held in Stockholm in May 2007 and convened by Banverket and Network Rail. Related work has examined automating the import of rail timetable data (Armstrong et al., 2007), whilst more general work has examined possible railway futures (Armstrong and Preston, 2007). Future work will examine how a hybrid system might be developed to assist in network train control.

Outputs:

Poster from RRUK Workshop 2008
Presentation from RRUK Workshop 2008
Armstrong, J., Preston, J. and Carlsson, M. (2007) Automating the Import of Electronic Timetable Data to EMME/2-Based Public Transport Models. International Association of Railway Operations Research Conference. Hannover, Germany.
Armstrong, J. and Preston, J. (2007) Possible Railway Futures. Presented at Transport – the Next 50 Years. Christchurch, New Zealand.
Balfe, N., Wilson, J.R., Sharples, S. & Clarke, T. (2008). Understanding Rail Signalling Automaton. Paper accepted for presentation at World Congress on Rail Research, Seoul, South Korea, May 2008.
Balfe, N., Wilson, J.R., Sharples, S. & Clarke, T. (2008). Structured Observations of Automation Use. Paper accepted for presentation at Ergonomics Society Conference, Nottingham, UK, April 2008.
Balfe, N., Wilson, J.R., Sharples, S. & Clarke, T. (2007). Analysis of current UK rail signalling systems. Paper presented at the Human Factors and Ergonomics Society European Chapter conference on Human Factors for Assistance and Automation, Braunschweig, Germany, October 2007.
Preston, J., Armstong, J., Bouch, C., Goodman, C., Weston, P. and Takagi, R.(2007) Decision Support System for Dynamic Re-Scheduling of Trains under Disturbance. 11th World Conference on Transport Research. Berkeley, California.