Dynamic Difficulty Level Adjustment Based on Score and Physiological Signal Feedback in the Robot-Assisted Rehabilitation System, RehabRoby

Abstract

In this letter, subjects take part in a fruit picker game on the robot-assisted rehabilitation system RehabRoby. The difficulty level of the game is adjusted using three feedback strategies. We use the scores of the subjects as the first feedback to determine the difficulty level in score feedback adjustment (SFA). We then consider physiological feedback, which is the mean skin conductance of the subjects, in physiological feedback adjustment (PFA). Finally, we use both the score and mean skin conductance to determine the difficulty level in combined performance and physiological feedback adjustment (CFA). Subsequently, we seek to understand the subjects' level of excitement by analyzing the subjective ratings using a survey, and by observing the statistically significant features generated from the collected physiological signals when they were performing the task using the proposed three feedback strategies. Survey results indicate that subjects respond positively to SFA and CFA but respond negatively to PFA because PFA proposes underchallenging levels to prevent a stressful experience, and as a result the subjects become bored. We find the skin conductance response (SCR), mean skin conductance (Mean <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">SC</sub> ), heart rate (HR), inter-beat interval (IBI), and mean temperature (Mean <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">temp</sub> ) as significant features between SFA, PFA, and CFA. The statistically significant features indicate that CFA proposes unpredictable difficulty levels, which keep the subjects more active, focused, and excited when compared to SFA and PFA.