Loading...

Developing a Sign Language-based Human-Robot Interaction of a Humanoid Robot

Hosseini, Ramezan | 2018

432 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 51390 (08)
  4. University: Sharif University of Technology
  5. Department: Mechanical Engineering
  6. Advisor(s): Meghdari, Ali; Alemi, Minoo
  7. Abstract:
  8. The statistics show that around 5% of the world's population suffers from deafness. For these people, the most comprehensive way of communicating with the world around them is using sign language. These facts will clarify the need for review in design practices, especially in designing user interfaces, in order to make technology usable for people with disabilities. The present study wants to design a sign language-based user interface for the RASA robot. The goal is to enable the robot to interact with deaf users. In the first step, a cognitive architecture developed, to enable the robot to interact by sign language in specific working conditions. The second step was the implementation of a sign detection unit. This unit, which uses the Hidden Markov Models, must recognize the signs performed by the user in a real-time procedure. 36 series of models were trained for the recognition task, and eventually, the best of them was chosen for final implementation. The chosen model achieved 84.5% overall accuracy in offline recognition of 14 signs. The final step was designing and implementing the interaction module. To evaluate the proper interaction in this section, a sign language-based two-player computer game was designed for the robot and the user to play together
  9. Keywords:
  10. Human Robot Interaction (HRI) ; Sign Language ; Sign Language Recognition ; Hidden Markov Model ; Statistical Pattern Recognition

 Digital Object List

 Bookmark

No TOC