Multimodal User Interfaces

With the rise of AI based models to recognize gestures, voice commands, biological signals, etc, we have become accustomed to using technologies such as voice virtual assistants (e.g. Amazon Echo) or gesture/face/object recognition software to interact with multiple devices such as unlocking your smartphone or changing the music being played. Whilst such modalities may be effective for certain tasks, how could they be used in other contexts? More specifically, how could multimodal interfaces (MMIs) which coordinate natural input modalities (such as voice, gesture, facial expressions etc.) with multimedia output modalities (such as video, audio, information visualizations, 3D graphics etc) unlock new potential forms of interaction?

This course is oriented towards Master students who want to gain a deeper understanding of the domain and will give them the basic techniques for designing, implementing and evaluating multimodal interfaces, as well as theoretical knowledge on multimodality, representation, visualisation of information and cognitive ergonomics.

Additionally, over the course of the semester, students (individual or group depending on the number of students) will have to design, develop and evaluate their own multimodal interface.

Evaluation Criteria

  • Written exam on the content seen during the courses (example questions are provided throughout the semester)
  • Individual or group project:
    • Final Report
    • Presentation(s)

Learning Objectives

By the end of the course, a student should:

  • Be able to design, program and evaluate a multimodal interface, thanks to the hands-on practice throughout the course.
  • Know the properties of multimodal interfaces.
  • Master the different levels of multimodal fusion.
  • Be familiar with the architectures of software adapted to multimodal interfaces and be able to understand synchronicity problems.
  • Have theoretical and practical knowledge of: gesture recognition, speech recognition, information visualisation and tangible interfaces.
  • Understand and know how to apply the various methods to evaluate multimodal interfaces.