COSMIC (Continuous Microgesture Acquisition and Sensing for Natural Interaction)
1 January 2026-31 December 2029
Instrument: WEAVE project under Grant no. 40035986
Funding body: FNRS & Luxembourg National Research Fund (FNR)
Area: SEN-3 Exact and Natural Sciences - 3
Keywords: gesture interaction, microgestures, radar-based sensing, deep learning, geometric algebra, gesture reocgnizers
Partners:
Université catholique de Louvain: Jean Vanderdonckt (PI), Sébastien Lambot (Co-PI), Nuwan Attygalle (senior researcher), Paolo Roselli (scientific collaborator)
Université du Luxembourg: Luis Leiva
Abstract
Although gesture-based interfaces have recently seen significant advances, current technologies remain limited by their reliance on discrete recognition gestures, constrained contexts, and a lack of continuous interaction. COSMIC will develop robust, continuous sensing techniques for real-time microgesture recognition, bridging the gap between expressive interaction and ecological validity. Microgestures are subtle, fine-grained hand movements performed with minimal effort, often while holding objects or interacting with surrounding environments. By leveraging multimodal sensor technologies such as wearable input devices, miniaturized mmWave radars, and emerging on-skin conductive materials, COSMIC will explore how microgestures can be detected in real-time, interpreted continuously, and utilized in natural settings.
The project follows a multidisciplinary, multi-stage approach, starting with the formalization of a comprehensive taxonomy of microgestures for real-world use cases (WP1). This informs the development of novel recognition algorithms, combining deep learning models with data-efficient analytic techniques (WP2). To support interaction in diverse environments, we will iteratively design and develop prototypes (WP3) that will be assessed through experimental testing with users in realistic scenarios (WP4). COSMIC has the potential to open new avenues for more intuitive, seamless interaction across domains such as wearable computing, ambient intelligence, augmented reality, and assistive technologies. The research team brings together leading expertise in gestureinteraction, radar-based sensing, machine learning, and Human-Computer Interaction. Ultimately, the project envisions a future where interacting with digital systems becomes as seamless and effortless as gesturing naturally in air.
Data Management Plan: URL
Ethical aspects
The project involves activities with human participants, notably during gesture elicitation studies (WP1) and user evaluations (WP4). As such, we will ensure compliance with applicable institutional, national, and international ethical standards.
- Informed consent: All participants will be informed about the study's objectives, procedures, and any potential risks or discomforts. They will provide consent using written forms.
- Data protection: Personal data will be anonymized or pseudonymized where appropriate. All data collection, storage, and processing will adhere to GDPR principles. Participants will retain the right to access, rectify, or erase their data at any point.
- Minimization of risks and discomfort: Our studies are low-risk by design. However, any physical or cognitive discomfort will be minimized through task breaks. Participants may withdraw at any time without consequence.
- Vulnerable populations: We will not involve vulnerable groups (e.g., minors, individuals with disabilities), only healthy participants.
- Ethics review: Before starting any studies involving human subjects, we will submit detailed protocols to our ethics committees at UCLouvain and Uni.Lu.
- Open Science: Datasets and code will be shared only in anonymized and ethically compliant forms.