Novel method of propulsion pattern recognition in a manual wheelchair simulator

Propulsion pattern recognition in a manual wheelchair (MWC) simulator contributes to better identify the users’ propulsion techniques. It can provide them with appropriate feedback and training, in order to prevent chronic shoulder pain. Recent work in our lab has focused on the development of an affordable wheelchair simulator, which provides force feedback (gravity, inertia) during propulsion in a virtual scenario, as well as visual feedback on propulsion performance. We now aim to add important feedback information about propulsion style, using a low-cost approach. Objective. The first objective will be to track and find the coordinates of the user’s wrist in a recorded video. The second objective will be to classify the set of wrist coordinates, which is associated with one push cycle, to one out of four possible patterns. Methods. We will use a simple webcam to record users from the side view, while they propel the pushrim. Then, by using the open-source DeepLabCut library, we will track and extract the wrist position.

Faculty Supervisor:

Philippe Archambault

Student:

Partner:

Université Paris-Saclay

Discipline:

Physics

Sector:

Education

University:

McGill University

Program: