Machine Learning Prediction on User Satisfaction in Human-Robot Interaction (HRI) Tasks

This research study investigates machine learning techniques to predict user satisfaction in Human-Robot Interaction (HRI) while performing teleoperation tasks using a mobile robot moved via rudder pedals. In these tasks, user satisfaction is the degree to which the user experience meets the user’s needs and expectations, including the level of comfort and acceptability perceived during interaction with the system. To collect the data to train machine learning algorithms, an experiment was designed involving 30 participants to analyze their task performance supported using a haptic glove. Two experimental sessions using and not the glove were done. So, experimental data was included in the database named Robot Motion Dataset (RMD). The database was then split into internal and external data sets. The first data set was used to train shallow machine learning algorithms based on Decision Trees, Support Vector Machines, Artificial Neural Networks, and so forth. Thus, the external data set was used to evaluate the performance of trained models. The best models achieved a level of accuracy of over 80% on the external data set. The research finding underscores the potential of data-driven, intelligent systems to provide real-time insights into the quality of HRI, also using the haptic glove. Hence, intelligent models can support operators or supervisors without interrupting the task and, at the same time, further reduce the risk factor.