HTO Analysis on Remote Shunting Operations. Final report within the framework of SBB Demonstrator Remote Driving

Kurzfassung

As research and development in the field of ATO progresses, the concept of remote monitoring and control of trains as a fallback level or as a regular solution for defined applications is coming into focus. In order to ensure the efficient and safe operation of such future systems, questions regarding the interaction of human, technology and organisation (HTO) must also be answered during the development phase. As part of an SBB demonstrator on remote train driving, the DLR Institute of Transport Systems Technology was commissioned by SBB to carry out an HTO analysis to determine user requirements and evaluate a prototype workstation for remote locomotive control in the context of shunting. To this end, requirements were researched from the literature and a field test was jointly designed and carried out to test the system. In the literature research, HTO requirements in the categories of sensor technology and technical infrastructure, workplace design, task design, individual requirements and operational and organisational framework conditions were excerpted from 36 identified sources relating to remote control in the rail sector and other domains and organised by category. To further specify the requirements with regard to human-technology interaction, a prototype workstation for remote shunting developed by Alstom on behalf of SBB and situated at the Alstom office in Zurich-Oerlikon was tested in the field. From there, an Aem 940 locomotive (Alstom Prima H4) was controlled at the Zurich-Mülligen shunting yard. During the test phase from 5 February to 31 March 2024, a total of 36 test sessions took place with a random sample of 24 SBB train drivers (TD) from the passenger, goods and infrastructure divisions. Tests were carried out during the day and at night to vary the ambient brightness. The participating drivers tested the system by working on 12 different scenarios relevant to shunting operations. These included starting the drive with different positions of the locomotive relative to the relevant dwarf signal, detecting various objects in and at the track, manoeuvring in small spaces and stopping precisely. Aspects of the effectiveness and efficiency of task performance and user satisfaction were recorded as dependent factors. The quantitative analyses were supplemented by comprehensive qualitative surveys, including on the nature of perceived challenges and possible solutions for system design as well as on the perception of possible changes to the job description of the driver. The majority of the shunting tasks could be carried out effectively with the tested system, with two exceptions: detecting a brake shoe on the track and determining the exact location of the vehicle in the shunting yard. The observed efficiency of task performance was reduced compared to shunting on the locomotive, as both the perceived time required and the subjective effort to achieve the results were higher. This was mainly due to system-related limitations in human perception, primarily at the level of vision, but also at the level of body perception and, to some extent, hearing. Differences in the system with regard to the lever control dynamics that drivers were used to also played a small part in reducing the efficiency of task execution. On the upside, particularly the extended video display of the Berne area in the system was emphasised as helpful, while the panoramic display of the video stream with wide peripheral areas and the design of the controls, which was largely in line with experience and expectations, were also perceived as positive. User satisfaction with the system design was in the medium range. In a further developed system for manual remote control in shunting operations, the visible areas should be enlarged compared to the tested system and the recognisability of objects in relevant areas should be increased. The remote operators (RO) require access to other vehicle functions (e.g. lighting) and a system that allows them to easily determine their own exact location in the shunting yard. Feedback on the quality of the data connection should remain part of the system and its functionality can be expanded to provide further support in dealing with latency. Further research and development is required to answer other design questions in the area of remote train monitoring and control. This concerns, among other things the question of the effects of the size and variability of latency for the purpose of a meaningful definition of critical values, the questions of how system design can support the handling of latency, which areas must be visible and in what quality, and which system design can best combine the coverage of information requirements with the capacities for data transmission, and the question of how the perception of action-relevant parameters such as speed, distances, traction conditions and gradients can be supported at the remote workstation. Further questions concern the effect of longer interaction experience and training with the system on performance as well as the design of the organisation around the workplace, including the description of new job profiles in a highly automated railway system.