The objective is to devise techniques for the automatic and creative generation of complex movements in robots, such as choreographies, and is characterized by robot learning, human-robot interaction, cognitive processes of movement learning and creativity.
Programming movements in robots is one of the main issues in robotics. First of all, it is necessary to bridge the physical space – where actions have to be performed – with the motor control space. Moreover, robots are embodied, i.e. they are immersed in a physical environment and the information they can use to take their actions comes from their sensors. This implies that the sequence of actions composing a movement must satisfy physical and structural constraints, as well as aesthetic preferences, such as smoothness and naturality of moves and be suitable and safe according to the robot’s body.
In this context, architectures for programming robot movements based on human cognitive models for action learning will be studied. Techniques that can automatically generate complex movements, either with set goals or through autonomous creation, will be investigated so that robots can independently create a complex sequence of movements, such as a choreography.
Research outputs:
- Tools for the generation of movements and choreographies
- Proposals for mapping robot/human performance movements into computer-based representations
- Integration of Artificial Intelligence and Cognitive Science methods to define and execute robot movements in performing arts