User-Centered Development and Evaluation of an Adaptive and Modular Human-Machine Interface for the Remote Operation of Highly Automated Vehicles

Abstract

For both technical and legal reasons, highly automated vehicles (HAVs, SAE level 4) will require the support of a human operator when deployed on public roads. In order to make vehicle automation more robust while ensuring safe, effective and efficient operations, the human operator could be located remotely to provide assistance to an HAV from afar. While the development of automation technology is progressing, proposals for a user-centered human-machine interface (HMI) of the remote operator's workplace specifically for remote assistance tasks remain scarce. The work presented here investigated novel human-machine interfaces (HMIs) for the remote assistance of HAVs with human-centered design methods. To the knowledge of the author, it presents the first systematic evaluation of a prototype of a workplace HMI for the remote assistance of HAVs and the examination of HMI features to support the remote operator in providing remote assistance that can be integrated into the prototype. Study participants' ratings of the prototype indicate satisfactory to good usability and above-average situation awareness while ratings for workload, albeit affected by a secondary task, did not exceed a medium degree. The prototype received medium ratings for user experience. Participants were able to complete the remote assistance tasks in all studies. Regarding two HMI features that were specifically examined, the augmented HMI for detecting relevant road users improved operator performance and SA and reduced mental workload, particularly in low-visibility situations. No major effects on these variables were found for alternative camera perspectives instead of the commonly used first-person perspective. Particularly the augmented HMI was shown to be able to compensate for unfavorable weather conditions regarding the detection of relevant road users by the remote operator: collisions with other road users and mental workload of the operator were significantly reduced while situation awareness was elevated. Further research could create contingency plans between requests for remote assistance, operator tasks, and HMI features to support the operator, requiring a deeper understanding of the interplay between tasks and suitable HMI features. Ultimately, contingency plans can inform an adaptive system for workplace HMIs to provide remote assistance triggered by context factors or the operator state.