Embedded Files
Reframing machine learning as a source of play a wearable exergame probe co-designed with children with and without Sensory-Based Motor Disorder
Reframing machine learning as a source of play a wearable exergame probe co-designed with children with and without Sensory-Based Motor Disorder
Overview
Overview
Cirkus is a wearable-technology design probe supporting many types of social, animal-locomotion exergames moving like kangaroos, lizards, crabs targeting strength, flexibility, movement planning, coordination, and endurance for children with Sensory-Based Motor Disorders (which often co-occur with autism). The flexible app lets children create their own games (a published catalog of 17 co-created exergames), and while they move, it collects accelerometer data that further trains a bespoke machine-learning movement classifier. The project deliberately reframes ML's role: rather than a diagnostic tool whose only success metric is accuracy, ML becomes a participatory, motivating material for co-designing therapy games and a teaching surface for digital literacy about how AI is trained and operates.
Problem & Motivation
Problem & Motivation
The historical success metric for an ML model is predictive power but a gold-standard accuracy benchmark can fail, and cause harm, when it inappropriately misrepresents a disabled or minority body. For children with SBMD, "messy" movement data isn't noise to clean up; it's the user. Cirkus asks what assistive ML looks like when accuracy stops being the only goal, and motivation, expression, and co-creation become first-class success criteria.
Approach, Team & Publications
Approach, Team & Publications
5 participatory workshops with 30 children with and without SBMD were conducted at the largest circus company in Sweden, a local school, and a lab setting (during Dr. Duval's visiting scholar stint at Uppsala University). Result: a catalog of 17 co-created exergames and a trained ML activity classifier. PI: Jared Duval | Visiting Scholar host (Uppsala): Annika Waern | Co-authors: Laia Turmo Vidal, Elena Mrquez Segura, Yinchu Li, Annika Waern | Publications: Duval, J., et al. Reimagining Machine Learning's Role in Assistive Technology by Co-Designing Exergames with Children Using a Participatory ML Design Probe. ACM ASSETS 2023. https://dl.acm.org/doi/abs/10.1145/3597638.3608421 | Turmo Vidal, L., Duval, J. Ambiguity as a Resource to Design for a Plurality of Bodies. Halfway to the Future 2024. https://doi.org/10.1145/3686169.3686176
Page updated
Google Sites
Report abuse