ECALAM - Embodied conversational agents for older people or people in vulnerable situations
Business, Management and Services
As populations age and increasing numbers of people experience a loss of independence, assistive technologies offer a major opportunity to support independent living at home and ease the burden on family caregivers. While current voice assistants such as ChatGPT, Gemini, Siri and Alexa have already demonstrated the value of natural-language interaction, they remain limited when it comes to complex, personalised uses in care settings.
This project explores the potential of Large Action Models (LAMs), a new generation of artificial intelligence capable not only of engaging in dialogue, but also of taking concrete action in digital environments—for example, scheduling an appointment, ordering a meal or contacting a relative.
A key innovation lies in integrating these models into Embodied Conversational Agents (ECAs), represented by on-screen avatars. Unlike interfaces based solely on voice or text, these visual agents can express emotions, use gestures to complement conversations and enhance the sense of presence. This makes interactions feel more natural and empathetic for users.
The aim is to assess the technical feasibility of these solutions, their relevance to everyday life—including reminders, social interaction and coordination with service providers—and their acceptability to the people concerned. The project also examines the ethical and societal issues associated with their use.
Conducted in collaboration with several care institutions in the canton of Fribourg and Pro Senectute, this work represents a first step towards developing intelligent assistants—or even companion robots—capable of providing practical support for people’s independence amid significant demographic change.
This project is funded by the HES-SO PEP Digitalisation.