ASSESSMENT OF TRAIN DRIVERS' PERFORMANCE IN A DRIVING SIMULATOR

In recent years, driving simulation has become one of the most effective methods for the training of train drivers. Besides its application for training, driving simulation is increasingly used as a research tool to study rail human factors. Nevertheless, the training and/or research activities rarely take advantage of all possible capabilities of the used simulator technology. Especially, the capabilities of behaviour and performance assessment by analysing simulator data remain unused. In order to face this deficiency, an integrative data concept (PERMA concept) for the assessment of train drivers' performance is presented that combines the objective data provided by the simulation with the observer's subjective ratings. This concept is firstly based on the detailed description and definition of the target behaviour derived from directives and regulations. Secondly, it specifies the in-depth analysis of the actual behaviour during specific simulator events. Finally, the actual and the target behaviour are compared using pre-defined thresholds for assessing the performance. This procedure allows an impartial, transparent, and detailed assessment, as it builds on and expands the several benefits of simulation as a training and research method for train drivers: Behaviour can be measured in a wide range of situations including events that cannot be reproduced in reality (e.g. equipment failures) and are very rare during daily operation (out-of-course events). Additionally, the events used for assessment can be reproduced and allow a standardised assessment for every train driver. If these benefits are further expanded by behaviour recordings, data storage, and assessment tools, a (semi-) automatic and objective comparison of the train drivers' performance against pre-set standards and criteria can be achieved. The paper presents a process model (PERMA model) for the two major steps of driver performance assessment, i.e. (1) the specification of exercise and assessment and (2) the assessment algorithm and the execution of the assessment. By that, the model explains the definition of assessment parameters as well as their processing and interpretation. Experiences at Deutsche Bahn show the applicability of the concept and prove its appropriateness for efficient research and training.