Development of robots that promote inclusion and diversity

It is essential to use representative data when designing social robots, says Ginevra Castellano, professor of intelligent interactive systems.

It is essential to use representative data when designing social robots, says Ginevra Castellano, professor of intelligent interactive systems.

How do we develop social robots that make everyone feel included and represented? This question is the focus of researchers Ginevra Castellano and Katie Winkle and also of an international conference to be held in Stockholm in March.

Professor Ginevra Castellano at the Department of
Information Technology. Photo: Mikael Wallerstedt

“If we are not careful about the data we use to build systems, we will end up with technology that does not work,” says Ginevra Castellano, professor of intelligent interactive systems.

Artificial intelligence has made it possible to develop robots that interact with humans and respond to human needs. At the same time, according to Ginevra Castellano, director of the Uppsala Social Robotics Lab, responsible AI use is essential if it is to have a positive impact on society. 

“An important part of creating ethical and reliable artificial intelligence and robotics is to ensure that they are inclusive and aim to reduce discrimination and bias. To do this, we need to include everyone in the design process, especially those groups or users who may be underrepresented due to gender, ethnicity, religion, culture or age, among other factors.” 

Risk of bias

Assistant Professor Katie Winkle at the Department
of Information Technology. Photo: Private 

Facial recognition is an example of a system that can complicate and do more harm than good if it is not built on representative data. This according to colleague and assistant professor Katie Winkle, an engineer with a PhD in robotics.

“The technology can get very good at identifying only certain types of people, which can lead to unfairness, discrimination, even exclusion from decision-making processes. That's why it's important that we address these challenges now," says Katie Winkle.

Support within healthcare

At the Uppsala Social Robotics Lab, robots are being designed and trained for use in areas such as education and healthcare. Currently, researchers are investigating how social robots can assist in screening women for signs of postpartum depression. The multidisciplinary project is based on a collaboration with psychiatrists, experts in perinatal depression and gender scientists, among others.

“When designing this type of robotic technology, it is important that we don't just think about the needs of the specific user. It's also about the societal norms surrounding depression and motherhood, which are important key elements to take into account in the design," says Ginevra Castellano.

Major international research conference

The research team's project will be presented at the world's largest conference in human-robot interaction, HRI 2023, in Stockholm between 13 and 16 March. The conference will be held in Sweden for the first time in its 18-year existence and will bring together international researchers in fields such as artificial intelligence, design, psychology and social sciences.

Ginevra Castellano is the conference chair and Katie Winkle is one of three local organisation chairs. Katie Winkle has been nominated for a best paper award at the conference for her paper "Feminist human-robot interaction: principles and practices for a better and more ethical human-robot relationship".

What do you hope the conference will achieve?
“We hope to raise awareness on the topic and the need to address human-robot interaction for all. And for more people to know about the good research being done in Sweden and Uppsala in human-robot interaction,” says Ginevra Castellano.

Anneli Björkman

Subscribe to the Uppsala University newsletter