The paper presents a long-term assessment of a service robot, Sacarino robot, in a hotel environment. Its aim is to act as a bellboy to greet guests and provide a few basic services to them, such as guiding them to a bar, providing weather information, and calling a taxi. Three-stage assessments have been implemented: stage 1, to set up and study on proxemics; stage 2, to implement metrics assessments; and stage 3, to refine the robot and reassess metrics, based on stage 2. The aspects to assess include the automation of the robot, the human-robot interaction (HRI), and the maintenance reliability. A popular robotic operation system (ROS) has been applied to address the robot automation with respect to path planning, autonomous charging, gesture control, and so on. A few HRI means were provided on the robot, such as a touchscreen, voice dialogs, gestures, and leading/following the guest. A state machine was designed to provide a logic relationship between these tasks.

This paper is a useful resource for researchers or enterprises that are willing to step into the service robot field. It provides many practical design considerations to take into account while also showing lessons to avoid. For example, the robot must be fully autonomous and have robust navigation. The involvement of manual operation should be reduced to a minimal level for nontechnical people to operate. Voice interaction could be improved with a “press to speak” button that the guest can press to start a conversation. Although it is inconvenient, it improved the recognition accuracy. Statistical information will be informative to a service robot designer, including the most popular service provided by the robot, the busiest time for the robot, how to deal with active kids, how to cooperate with hotel staff, and how to treat the misuse of the emergency stop function. From stage 2 to stage 3, the authors improved the robot in several aspects in order to increase the percentage of HRI in the operation. They all look meaningful.

As the paper focuses mainly on a performance assessment in a human environment, it didn’t give too many details of the technology behind it. It indicates a few challenges that should be solved in the future, such as voice recognition in a noisy area, a dynamic navigation algorithm in a changing environment, and the robustness of the hardware and the HRI in different context and subjects.

Regarding the paper itself, the quality of the figures could be improved. Some flow charts are misleading, and captions are not fully self-sustained. However, it doesn’t affect the paper’s contribution.

Overall, this paper is well written, and the work is technically sound for researchers of service robots.