Smart Glasses
What we want to find out
This study aims to explore how computers can better understand social interactions. We are combining smart glasses (a type of wearable technology) with artificial intelligence, using a data analytic approach called machine learning (ML). We are investigating if ML can accurately recognise and interpret people’s thoughts and feelings (mental states) in real time. We are also exploring whether this technology can provide real-time and useful feedback during conversations to support understanding of social cues, making social interactions easier and more accessible.
Why we’re doing this research
Many people find social situations challenging and confusing. We want this research to improve how technology, such as machine learning, understands social scenarios and identifies mental states so it can offer useful support to people that would like support during social interactions.
Method
This project will first examine and test the ability of state-of-the-art ML algorithms to identify mental states using existing audio and video databases of actors expressing different emotions. Furthermore, we are working with a company called Emteq Labs, who have developed smart glasses that are capable of detecting signals linked to facial expressions and emotional responses. This will allow us to explore how this technology could work in real-life situations.
Community engagement
Autistic people will be closely involved throughout the project. We are working with the autistic community through ongoing consultation and collaboration with members of CARD. We will seek feedback on the study design, language, relevance, and usability of the new assessment tools to ensure they are accessible, respectful, and aligned with the goals of the autism community, reinforcing the project’s commitment to co-designed research and neurodiversity-affirming practice. This project has received additional funding from the Cambridge Public Engagement Starter Fund to conduct focus groups throughout the design state of this project.
Potential impact
This study could illustrate how combining wearable technology with advanced ML can provide support for people that experience differences in social interaction. We also aim to emphasise the preferences of the autistic community in the development of this technology. This project will apply particular focus on the development of neuroaffirmative AI. Lastly, this project aims to advance the scientific understanding how people interpret social information by developing computer models that can identify thoughts and feelings, even when the data are unclear.
Results
We anticipate to share results in October 2028. Updates will be posted here.