What is the Role of Anthropomorphism When Robots Make Mistakes? Trading off Anthropomorphism in Imperfect Human-Robot Collaboration

Service robots, designed to perform useful tasks and serve as companions, are being increasingly applied in many fields. However, whether anthropomorphism can enhance human performance in collaborative tasks or might turn into a negative impact when the error rate is also on the rise remains unknown. Our study used hand-over-hand tasks performed at varying collaboration success rates, recording parameters of collaborative movements with different anthropomorphism robotic hands in real time. The results revealed an interaction effect between robot performance and anthropomorphism and the match of them may not fit a linear model. Our conclusions provide suggestions for the design of the service robot's appearance. It is recommended that robot anthropomorphism should be determined according to its potential performance. When performance is optimal, anthropomorphism should be increased; conversely, when performance is suboptimal, anthropomorphism should be reduced. This approach may prove an effective method to promote the performance of human collaborative actions and human-robot collaboration.