2021-2022 Master Projects
Investigating the Link between Human’s Mental Model and Robot’s Explainability
This project aims to contribute to the question “Can robot’s ability to understand human’s perspective contribute to its ability to explain itself or its decision when needed?” The explainability of embodied agents is defined as their ability to provide information about their inner workings using social cues such that an observer can infer how and why the embodied agent behave the way it doesˆ. There are several definitions or mechanisms involved in the way explainability is used or investigated in the literature. On the other hand, perspective taking refers to the ability to view a situation from the perspective of others and anticipate their intention and it is fundamental to becoming competent intellectually and socially. Taking someone else’s perspective can be on a perceptual, cognitive, or affective dimension. Humans use their ability to take or infer others’ perspective daily over each dimension without being aware of it.
While developing Autonomous Intelligent Systems (AIS) and robots, one of the key design factors is making them expressive and socially aware beside improving their transparency and establishing trust. All the efforts considered, humans still make their own mental models of the robots and its abilities based on their previous interactions with living things. This can result in incorrect or inadequate mental models of the robot or agent that can result in self-deception and distrust in the system*. In this project we aim to investigate how the robot’s ability to understand the human’s mental model of the robot can help the robot to explain itself either in relation to the task or to correct an inaccurate mental model. This study feeds into a more comprehensive framework that tends to model the robot’s cognitive abilities in understanding and modelling human perspective and mental models.
Literature:
ˆWallkötter, S., Tulli, S., Castellano, G., Paiva, A., & Chetouani, M. (2021). Explainable Embodied Agents Through Social Cues: A Review. ACM Transactions on Human-Robot Interaction.
* De Graaf, M. M., & Malle, B. F. (2017, October). How people explain action (and autonomous intelligent systems should too). In 2017 AAAI Fall Symposium Series.
Keywords: Robotics, Theory of Mind, Cognitive Modelling, Explainability, Human-Robot Interaction
Developing Modalities for Social Robots to Promote Prosocial Behaviours
Perspective taking; the ability to view a situation from the perspective of others and anticipate their intention, is fundamental to becoming competent intellectually and socially. This ability plays a crucial role in adapting and improving social and interpersonal interactions. In recent years a growing body of research has focused on developing this ability in embodied and virtual agents. Prosocial behaviours refer to voluntary acts intended to benefit or help others, and according to Computers Are Social Actors (CASA) framework, humans mindlessly apply the same social heuristics to computers as they apply to humansˆ.
The project investigates the implications of robot exhibiting perspective taking abilities in fostering the human’s prosocial behavior toward the robot. The development of the robot’s perspective taking abilities consist of using cognitive models and machine learning algorithms to train the robot’s abilities to perceive and predict human’s intention within the interaction. On evaluating the human’s behavior toward the robots, the optimal research plan includes investigating different functions of prosociality (proactive, reactive, and altruistic*). The context of interaction between human and the robot and the type of robot used is open for discussion.
Literature:
ˆOliveira, R., Arriaga, P., Santos, F. P., Mascarenhas, S., & Paiva, A. (2021). Towards prosocial design: A scoping review of the use of robots and virtual agents to trigger prosocial behaviour. Computers in Human Behavior, 114, 106547.
*Danielle Findley-Van Nostrand & Tiina Ojanen (2018) Forms of Prosocial Behaviors are Differentially Linked to Social Goals and Peer Status in Adolescents, The Journal of Genetic Psychology, 179:6, 329-342, DOI: 10.1080/00221325.2018.1518894
Keywords: Robotics, Perspective Taking, Prosocial Behaviour, Machine Learning, Human-Robot Interaction
