Titelaufnahme

As robots are increasingly capable of working alongside humans, human trust in robots has emerged as a prominent research subject. Based on psychological theories of trust formation and findings from related scientific fields, robot understandability and predictability have been repeatedly discussed as essential factors for developing cognitive trust in robots. Though, questions remain regarding the empirical applicability of these findings to human-robot interactions and on how to design industrial robots to act more understandably and predictably. To systematically address these questions, a randomized active-control trial was conducted. In a virtual workspace, 124 participants played a self-developed research game, requiring them to predict the goal objects of a mobile manipulator’s reaching movements. Drawing on kinematic research findings, the understandability and predictability of the robot’s reaching movements were varied through the positioning of informative object alignment cues. More precisely, participants either saw the robot align its base joint and gripper (i) throughout its trajectory (active control), or it primarily aligned its (ii) base joint, (iii) gripper, or (iiii) base joint then gripper to its goal objects. Results show a significant superiority in understandability and earliness of prediction for trajectory groups with a preceding gripper and base joint cue. Mediation analyses further revealed that higher understandability ratings positively affected participants’ predictability perceptions but not their trust in the robot. This study sheds light on the relationship of the communicational factors proposed to underly cognitive trust in robots. Moreover, it introduces a novel approach for shaping industrial robots’ movements to be more understandable and suitable for effective interactions with humans.