 |
COIN consists of two universities and four departments in Sweden. The multi-disciplinary aspect of the project requires for a multi-disciplinary group.
|
|
Role: The Division of Decision and Control Systems (along with the Division of Robotics, Perception and Learning) at KTH will
firstly be investigating the creation of an abstract model of the robotic system in interaction with its environment (including humans). This model can then be exploited to synthesize a feedback controller for the robot to satisfy high-level specifications with a temporal logic formulation. In case of failure of this controller synthesis, co-adaptation methods will be developed in order for the robots and humans to adapt the abstract model or the specification until satisfaction of the updated control problem can be guaranteed.
then investigate the mix-initiative motion and task planning problem when the human intention is explicitly modeled in both the continuous controller and contingent high-level task assignment. Novel motion and task planning algorithms will be developed that ensured all-time safety, fulfillment of the contingent tasks and asymptotic learning of the human preference.
Personnel: Prof. Dimos V. Dimarogonas (PI), Sofie Andersson, Dionysis Theodosis, Péter Várnai, Pedro Pereira, Pedro Roque, Pedro Miraldo, Xiao Tan, Wenceslao Shaw Cortez
Role: The Social Robotics Lab, Department of Information Technology (along with Division of Decision and Control Systems at KTH) will be investigating (i) Long-term learning framework and co-adaptation for personalized autonomy; (ii) Affect-based co-adaptation approaches; (iii) Synthesis of guidelines for a human user; (iv) Long-term evaluation with human-in-the-loop experiments.
Simulating the tremendous social adaptation abilities that characterise human interactions requires the establishment of bidirectional processes in which humans and robots synchronise and adapt to each other in real-time by means of a mutual exchange of verbal and non-verbal behaviours in order to achieve mutual co-adaptation. To endow robots with co-adaptation abilities, we aim to develop approaches that leverage statistical learning to incrementally adapt robot’s behaviours and strategies to a specific user’s situation, for example, a user’s emotion, personality or progress with a task. This affect-based, personalised co-adaptation aims to close the human-machine loop while enabling robot learning from human users in more natural, intuitive ways.
Personnel: Prof. Ginevra Castellano (PI), Mohammad Obaid, Alex Yuan Gao, Giulia Perugia, Maike Paetzel
Role: The Division of Robotics, Perception and Learning at KTH (along with Division of Speech, Music and Hearing at KTH) will be investigating (i) Natural language interaction for planning and execution; (ii) User modelling framework; (iii) Adjustment of system behavior for short-term adaptation; and (iv) Entrainment of user behavior.
Adapting to an interaction partner requires the ability to anticipate what he or she is going to do in the near future. Imagine a handover scenario - the receiver needs to anticipate the location of the handover that is intended by the giver in order to adapt. Otherwise, the giver has to wait unnecessarily long before the receiver reaches this position. To make it easier for a partner to adapt, it is also of advantage to actively signal intentions. For example, the giver might fixate on the end location to indicate intentions. We address these concepts in the light of human-robot interaction from two viewpoints. On the one hand, we develop predictive models of human motion so that the robot can react to human intentions. On the other hand, we investigate how the state and plans of the robot can efficiently be communicated to the human partner.
Regarding the way of the more natural and efficient communication between robots and humans, we consider both verbal and non-verbal means of communication. The general idea is to create a framework that combines dialog, gestures, gaze direction and augmented reality so that even non-experts could operate robots and preform complex collaborative tasks.
Personnel: Prof. Danica Kragic (PI), Elena Sibirtseva, Judith Bütepage
Role: The Division of Speech, Music and Hearing (TMH) in the School of Electrical Engineering and Computer Science (along with Division of Robotics, Perception and Learning, and Division of Decision and Control Systems) at KTH will be investigating the usage of natural, unrestricted spoken interaction between humans and robots, with the goal of developing systems which can be “conversed with” rather than simply issued isolated commands. An important aspect of such interaction is that the robot will have to adapt its way of interacting to the human partner. By modelling the user, the robot will be able to adapt its language behaviour to the specific user, for example in terms of conversational style and vocabulary.
Personnel: Prof. Gabriel Skantze (PI), Erik Ekstedt, Nils Axelsson