AlterEgo · Enhancing Social Interaction with an AlterEgo Artificial Agent

The objective of AlterEgo is the creation of an interactive cognitive architecture, implementable in various artificial agents, allowing a continuous interaction with patients suffering from social disorders, by virtue of changes in behavioural (robot-based) as well as morphological (avatar-based) properties of that agent. The AlterEgo architecture is rooted in complex systems, machine learning, and computer vision. The project includes research in fundamental and clinical neurosciences, interaction modeling, development of new computer-vision techniques and human-robot interfaces, as well as evaluation of the scenarios with patients before, during, and after training sessions. At the end of the project, we will produce a new robotic-based clinical method able to enhance social interaction of patients. This new method will change existing therapies, will be applied to a variety of pathologies, and will be individualized to each patient. The first challenge of this project is to create an avatar and a robot alter ego of a patient. Based on recent work in social robotics and neurosciences, we hypothesize that if patients face artificial agents morphologically and behaviourally similar to them, they will increase their social interaction. The second challenge is to transform the morphology and the behaviour of the similar artificial agent into that of a healthy and different agent. We assume that the smooth and continuous behavioural shift from similar/unhealthy to complementary/healthy social behaviour will lead to a better social rehabilitation. AlterEgo opens the door to a new generation of social artificial agents in service robotics. AlterEgo is an interdisciplinary project composed of 7 workpackages, at the interaction of Social Motor Neurosciences (UM1), Robotics (EPFL), Complex Systems Dynamics (UOB and UNEXE), Computer Vision (DFKI), and Psychiatry (CHU).