Towards trustworthiness and transparency in social human-robot interaction

Cooperation between autonomous robots and humans is becoming more and more demanding. Robots have to be able to capable of possessing and expose a wide range of cognitive functions, once humans require their help. This paper describes a cognitive architecture for human-robot interaction that allows a robot to dynamically modulate its own level of social autonomy every time a human user delegates to it a task to accomplish in her/his place. The task adoption process leverages on multiple robot's cognitive capabilities (i.e. the ability to have a theory of mind of the user, to build a model of the world, to profile the user and to make an evaluation about its own skill trustworthiness for building the user's profile). On the basis of these capabilities the robot is able to adapt its own level of intelligent collaboration by adopting the task at the different levels of help defined in the theory of delegation and adoption conceived by Castelfranchi and Falcone. Besides that, the architecture enhances robot's behavior transparency because gives to it the ability to provide a comprehensive explanation of the strategy it has adopted for accomplishing the delegated task. We propose an implementation of the cognitive architecture based on JaCaMo framework, which provides support for implementing multi-agent systems and integrates different multi-agent programming dimensions.