XOGAMOS: Intelligent Interactions in Social Robots for Cognitive Stimulation Activities
The XOGAMOS project addresses the improvement and sustainability of cognitive stimulation therapies in the elderly through AI techniques and models that will be integrated into a social robot. From a user-centered approach, the project addresses the challenge of ensuring the reliability of the integration of machine learning models that tend to suffer the well-known hallucinations and that generally make them inadequate in the context of real use. Through an approach based on pipelines that allow reasoning and verification of the consistency and suitability of the content generated by artificial intelligence for cognitive activities, the aim is to be able to evaluate the capacity and value of incorporating the models that will be developed based on AI to support cognitive stimulation activities.
Objectives
Thus, this action aims at investigating theory, concepts, techniques and methods to safely incorporate models to deliver more efficient therapy in dialogic multimodal contexts for cognitive stimulation with a social robot. Key points for the project:
- Multimodal QAs: Multimodal models for (visual) question-answering and dialogues
- Social robot for cognitive stimulation: Dialogic activities ready for cognitive stimulation with a social robot.
- Responsible AI: Designing with user-centric and responsible AI principles in mind.
Project
/research/projects/interaccions-intelixentes-en-robots-sociais-para-actividades-de-estimulacion-cognitiva
<p>The XOGAMOS project addresses the improvement and sustainability of cognitive stimulation therapies in the elderly through AI techniques and models that will be integrated into a social robot. From a user-centered approach, the project addresses the challenge of ensuring the reliability of the integration of machine learning models that tend to suffer the well-known hallucinations and that generally make them inadequate in the context of real use. Through an approach based on pipelines that allow reasoning and verification of the consistency and suitability of the content generated by artificial intelligence for cognitive activities, the aim is to be able to evaluate the capacity and value of incorporating the models that will be developed based on AI to support cognitive stimulation activities. </p><p>Thus, this action aims at investigating theory, concepts, techniques and methods to safely incorporate models to deliver more efficient therapy in dialogic multimodal contexts for cognitive stimulation with a social robot. Key points for the project:</p><ul><li>Multimodal QAs: Multimodal models for (visual) question-answering and dialogues</li><li>Social robot for cognitive stimulation: Dialogic activities ready for cognitive stimulation with a social robot.</li><li>Responsible AI: Designing with user-centric and responsible AI principles in mind.</li></ul><p><br></p><p><br></p> - CNS2024-154915 - Alejandro Catalá Bolós
projects_en