AMI
The team’s research is organized along three axes: Movement and Image, which explores representation and utilization of movement for interaction; Humans and AI, focusing on topics such as machine teaching, co-agency, and the environmental impact of AI; and Disabilities, dedicated to designing tools for disability evaluation and assistance.
Several interconnected lines of research are developed in the team to foster embodied interactions through novel representations, models, and applications of human movement in human-computer interaction. First, AMIArchitectures et modèles pour l'Interaction contributes to developing efficient models for image, video, and human movement analysis using computer vision and machine learning. Key contributions include creating compact data representations for recognizing activities, gestures, and sign language, as well as anomaly detection in videos. Additionally, sensor-based movement analysis techniques are developed with hardware and software methods for human movement analysis. Second, the team designs and evaluates movement-based interactive systems, applying them in performing arts, interactive arts, and movement learning. Projects explore the sensory impacts of virtual body representations and interactive sonification. Finally, the team conducts fundamental research in cognitive science concerning the agency of movement and its importance for designing interactions with computers and robotic systems.
The AMIArchitectures et modèles pour l'Interaction team focuses on developing human-centered approaches to machine learning, mainly through Interactive Machine Learning (IML), which prioritizes user involvement in the learning process with rapid, incremental model updates. One goal is to promote collaboration between developers, machine learning experts, domain experts, and end users. The team develops software tools and studies how people interact with such technologies, depending on their expertise. Another line of study explores how the sense of agency can change during joint actions with AI in comparison with joint actions among humans. Finally, the team addresses the environmental impact of AI and digital technology, advocating for sustainable research practices.
Research in the “Disabilities” theme deals with designing and evaluating assistive technologies for people with special needs. Projects are carried out for various cognitive, sensory, and motor disorders. The team studies technologies dedicated to motor disability, including tactile interfaces for wheelchair control, that aim to reduce the effort for individuals with neuromuscular diseases and connected clothing and smart objects that assist musculoskeletal disorder rehabilitation. AMIArchitectures et modèles pour l'Interaction also contributes to establishing evidence-based frameworks for assistive technologies for neurodevelopmental disorders such as autism and ADHD, as well as methods based on eye tracking for analyzing visual and attentional patterns. Finally, research on accessibility for the visually impaired is conducted. In particular, the team conducts fundamental studies on the perception of 2D structures on tactile pin matrices to help users navigate complex digital documents.