Exploring NAO Robot Errors in Collaborative Work: An Assessment based on Identification, Analysis, and Troubleshooting

A laboratory practice session conducted by a robot NAO was designed and implemented as part of a human-robot collaborative task session. With this purpose, the robot provided the participants with a set of step-by-step instructions that had the objective to turn on a LED with an embedded device. However, this study had the aim to evaluate the Pratfall effect in a teaching assistant scenario, thus, the robot provided both correct and incorrect instructions in some moments of the practice session, and the participants were requested to follow the robot’s instructions. Afterward, their reactions and perceptions were analyzed. To measure the participants’ perception towards the robot the NARS, RoSas, and questions about the instructions with errors were applied to the participants to identify their perceptions towards a non-perfect robot. It was found that all the participants finished the session, 65% of the participants could turn on the LED at the end of the session by following the robot’s instructions, and 66% of the participants realized the presence of errors. Lastly, it was found that the participants tended to present negative emotions every time the robot committed a mistake.