UppTalk Weekly: Design of Social Intelligent Robots and their interaction with humans
Ginevra Castellano, professor of Intelligent Interactive Systems at Uppsala University and director of its Social Robotics Lab, discusses human-centered social robots with host Lina Sors Emilsso on UppTalk Weekly.
In this talk
- 2:55
What makes a robot social
Castellano describes intelligent interactive systems as robots that perceive their surroundings, interpret what is happening, and plan actions toward goals. Social robots often have humanlike features that support familiar cues such as looking, pointing, moving, or smiling. Their appearance alone is not enough: they also need to recognize people, interpret their behavior and intentions, and make their own intentions understandable.
- 7:34
Designing for human interaction
Castellano studies robots in socially assistive roles, including classroom instructors, hospital assistants, companions, and tools for therapy or rehabilitation. Her team observes how people work in a given setting and draws on human behavior when designing robot communication. For example, the Dorsa robot was developed to support children and teachers in classrooms.
- 14:24
Fit the robot to the people and setting
There is no single ideal level of human likeness. Very humanlike robots can be unsettling when their movement or behavior falls short of human expectations, a response known as the uncanny valley effect. Castellano says a robot’s design should depend on its purpose, context, and users; a simpler robot may work as well in some applications.
- 18:10
Autonomy, trust, and responsibility
Castellano says social robots currently have limited autonomy, while progress in AI is expanding what they can do without human input. She raises questions about oversight, intervention, and accountability when systems make mistakes. Trustworthy robots should address fairness, inclusion, bias, and transparency, with people involved throughout design and deployment.
- 24:11
Support people without displacing them
Castellano sees potential for robots to assist teachers with STEM learning and to support children who need help outside the classroom. In one example, a robot tutor can encourage a child who is struggling or adjust an activity to their needs. She also cautions that children and older adults may overestimate a robot’s abilities or form attachments, so social support should be designed with those risks in mind.
Briefs summarise talks and interviews; timestamps open the video at that moment. Claims in a talk are the speaker's own.