Movement-based Human-Machine Collaboration: a Human-centred AI approach (accreditation to supervise research)

The context of this thesis is the collaboration between humans and machines in various industrial real-world situations. I propose collaboration mechanisms that are based on Human-centred Artificial Intelligence, which I define as methods and concepts of machine learning and pattern recogni- tion on signals recorded from the human body. I am interested in enabling human-machine partnerships in which the machine can understand and an- ticipate the human gestures and actions and react accordingly. Two scientific and technological hypotheses have oriented my research in movement-based Human-Machine Collaboration: 1. whether a machine can learn to recognize kinematic parameters of situated expert and non-expert gestures; and 2. whether gesture recognition can be used as an alternative to instrumental interaction mechanisms. These hypotheses were confirmed through a number of tests and experiments conducted on Human-Robot Collaboration, computer-mediated sensori-motor human learning and Digital Musical Instruments