Usability of Anthropomorphic Sound Feedback for Service Robot Malfunction through Repeated Experiences - with Emphasis on Vacuum Cleaning Robots

The service robot has a complicated interface and therefore users may find it hard to understand the system and structure of the robot. Users’ misunderstanding of the system of the robot can cause malfunctions of the robot, which can bring users serious problems and unexpected danger. In order to reduce the damage from malfunctions, research on the feedback of malfunction is required. Moreover, a service robot can produce a novelty effect, the tendency for performance to initially improve when new technology is instituted due to increased interest in the new technology. For this reason an investigation of the usability of feedback through repeated serial malfunction is needed. The aim of this study is to apply the anthropomorphic elements to the feedback of malfunction for service robots and evaluate the usability of the anthropomorphic sound feedback for a service robot through repeated malfunction. The experiment was carried out with two types of feedback (anthropomorphic vs. nonanthropomorphic) to evaluate the usability of the anthropomorphic feedback. Sixty-seven university students or graduate students at an engineering college were recruited as subjects for the experiment because demographically users of vacuum cleaning robots were mostly in their twenties and accustomed to new technology. In the experiment, subjects were required to deal with repeated malfunctions. The usability scale is comprised of effectiveness, efficiency, and satisfaction, which was evaluated through a questionnaire. Participants’ behaviors were analyzed by protocol analysis, video analysis, and interviews. The results from the experiment are summarized into four parts: First, the more repeatedly a malfunction occurred, the more effectible the anthropomorphic feedback is than nonanthropomorphic feedback; second, the anthropomorphic feedback is more efficient than nonanthropomorphic feedback in similar cases; third, the anthropomorphic feedback is overall more satisfactory than non-anthropomorphic as the feedback of malfunction; fourth, anthropomorphic feedback helped relieve temper and reduce aggressive actions caused by stress. These findings will provide an essential basis for feedback design of service robot malfunction.