Adaptative and Personalized Robot Behavior for Cognitive Training

Antonio Andriella, Guillem AlenyĆ  and Carme Torras

Institut de Robòtica i Informàtica Industrial, CSIC-UPC, C/ Llorens i Artigas 4-6, 08028 Barcelona, Spain.

Abstract: Current research supports the use of cognitive games to improve and monitor the brain function of elder people affected by Alzheimer Disease (AD) or Mild Cognitive Impairment (MCI). In this work we present an adaptive robotic system that provides a personalized assistance to patient performing a SKT subtest through encouragements and suggestions using verbal and gestural communication. First, the robot interacts with the caregiver in order to set up the preferred levels of interaction for that specific patient. Then the robotic system uses an adaptive algorithm to: i) define at each step of the game the best suitable action of interaction to support the user and ii) personalize the engagement level based on the user behavior. We have conducted a preliminary validation of our adaptive algorithm in simulation and an implementation in a real scenario. Our experimental results show that the adaptive algorithm is consistent and perform as expected, providing the user with the most suited engagement actions during a cognitive game such as SKT.


An example interaction of the robotic system assisting a person while he is playing SKT game: