NOMA: Unlocking the World
Through Inclusive UI

How might we create affordable assistive solutions to interact with physical devices?

Artifical Intelligence
Human-Centered Design
Product Development
Project type: Product development of AI assistive technology
Duration: Sep '19 - ongoing
Collaborators: Olivia Seow & Aditi Kumar
Awards: MIT Design for Social Impact First Prize, MIT Machine Intelligence Community Generator First Prize, MIT $50k Collaborative Intelligence Entrepreneurship Finalist

Context

Assistive technology can be a great democratizer. Digital tools in particular like screen readers and voice assistants have helped reduce the digital gap for older adults.
There is still a long way to go when it comes to interacting with physical devices. Assistive devices in this space are prohibitively expensive and have limited uses.
We believe that conversational interfaces are the next frontier for accessibility & inclusion; particularly for users that value self-reliance over dependence.

Process

The idea for this concept was born out of a design sprint in collaboration with Perkins School of the Blind. Given the positive reception we were encouraged to expand our user research by conducting primary and secondary research to validate our concept. Primary research included user interviews, ethnography and usability studies. We also validated our technology with multiple experts in the field.
We prioritized creating a conversational interface to assist with interaction with home appliances as our beach head application.

Solution

Our solution is easy to set up, adapts to user expertise and enhances continuous gestural improvement. All the user needs to do is pull out their phone, set it up on provided stand and open the NOMA app. The user can then ask for guidance to perform tasks on their existing devices. NOMA will track hand gestures to ensure to calibrate their response.

How does it work?

Social & Ethical considerations

  • Risk of over-reliance: consider ensuring adaptation of verbal feedback based on user skill-level to avoid enhancing degradation in existing faculties.
  • Risk of over-confidence: must carefully calibrate feedback based on user research and adopt a safety-first approach to avoid externalities of users letting their guard down while using NOMA.
  • Risk of faulty generalization of our AI: progressively test and rollout additional use cases as well as collect instances of faulty generalization to improve model.

Future Work

At scale, this project aims to contribute to creating more inclusive products.
Project currently on pause due to COVID-19 restrictions .