Family companion robot

Innovative research design for understanding emotions through multimodal home interaction data analysis.

A person with an expressive face, possibly laughing or displaying a strong emotion, wearing a maroon shirt. The lighting is dim with a focus on the face and upper body against a black background.

Innovative Research Design Solutions

We specialize in multimodal interaction data analysis, integrating emotional insights and personalized user experiences to enhance family dynamics and emotional well-being.

A smiley face emoticon is seen hanging from a string, with a blurred background that has warm, golden tones. The emoticon has large, expressive eyes and an open mouth with a tongue sticking out.

A close-up of a person's face with expressions indicating intense emotion, possibly yelling or expressing strong feelings. The lighting is bright and the person's features, including a beard and open mouth, are prominently visible.

A close-up of a young child with curly hair, expressing a strong emotion. The child’s eyes are tightly closed, and the mouth is open wide, suggesting crying or shouting. The skin tone is soft, and the earring adds a subtle touch to the image.

Our Approach to Research

Our team focuses on building dynamic affective graphs, fine-tuning AI models, and conducting experiments to improve emotional understanding and user engagement.

Innovative Research Design

We specialize in multimodal data collection and emotional analysis for advanced research design solutions.

Data Collection

Collecting 500 hours of multimodal home interaction data with emotional labels and user feedback.

A group of children are enthusiastically raising their hands and making expressive faces, possibly during a game or an interactive activity. The scene is lively and crowded, with the children wearing colorful clothes. There is a sense of movement and excitement among the group.

Dynamic Graphs

Building personalized family affective graphs integrating psychology literature and user-specific emotional tags.

Fine-tuning GPT-4 for dialogue adaptation, cross-modal alignment, and long-term memory integration.

Model Deployment

A close-up of a person’s face with hands covering parts of their face. Blue eyes with a tear, rough stubble, and expressive features suggest emotion.

A black and white image capturing a blurred, emotional expression of a person with hands covering part of the face, creating a sense of movement and intensity.

Emotion Analysis

Integrating emotional labels with multimodal data for insightful research.

A humanoid robot with large, expressive eyes stands prominently, capturing attention in a tech event setting. People in business attire are visible in the background, engaged in conversation and examining displays. The scene is vibrant with a backdrop that features the phrase 'Connected All Over Europe' in bold pink lettering.

Family Graph

Creating dynamic family affective graphs using personalized emotional insights and established psychological literature to enhance understanding and improve emotional well-being.

A hand holds a smartphone displaying a screen where various cartoon emoji faces are featured. The background is softly blurred, indicating an indoor setting with possibly a shelf or furniture.

Model Testing

Comparative analysis of fine-tuned models against baselines focusing on emotion detection accuracy, user retention rates, and privacy considerations, ensuring robust evaluation methodologies.