URATech Sampling
Accessible Multimodal Data Collection Platform for Severe Speech and Motor Impairment (SSMI)
URATech (sampling.uratech.life) is a bilingual web application designed to gather a small number of personal examples—including face photos, facial gestures, eye gaze, and optional speech—to tailor assistive communication tools to each individual. By employing a few-shot learning approach, the platform ensures recording sessions remain short and low-effort for participants with progressive motor loss.
● The Challenge
AI technologies often lack representation from people with disabilities in their training datasets, resulting in systems that fail to recognize diverse physical abilities, mobility aids, or atypical speech. Collecting this necessary data is difficult because in-person studies present severe mobility and cost barriers, while traditional web surveys are inaccessible for users relying on eye-tracking or switch controls due to small hit targets and high cognitive fatigue.
● Our Solution
URATech provides an online infrastructure that enables scalable, remote data contributions directly from disability communities in a controlled, respectful manner. The platform is engineered specifically for individuals with SSMI, utilizing a highly accessible interface featuring large tap targets (64px or more), choice cards instead of free text, and full keyboard support. Furthermore, it preserves user privacy through edge inference, dropping raw video feeds locally and securely transmitting only anonymized facial mesh coordinates for the few-shot learning model.





