Project for the course of Biomechanics of Motion (Fall 2019) lectured in IST from the University of Lisbon. Programmed in Matlab.
This project focused on 2D analysis of human mechanical motion. Hence, human motions that occur mainly along a single 2D plane are chosen.
Two different motions were studied: a simple walk stride and a deadlift exercise.
| Walk stride example | Stick-figure constructed from the collected lab data |
 |
 |
| Deadlift example | Stick-figure constructed from the collected lab data |
 |
 |
In a laboratory environment, the subject is asked to walk in the direction of one of the main axis of the laboratory, so that all points can be projected onto a known plane for 2D analysis of the movement. In practice, this is very hard to achieve and small deviations always happen. To correct for this, we use PCA (Principal Component Analysis) to detect the main direction of the movement and then project it back onto a known plane.
In this example, PCA is used to force the initial movement (in red) to happen in the X direction (in blue):
The benefits of this correction are clear when both versions (corrected and non-corrected) of the gait analysis are compared below.
At least two big artifacts observed in the non-corrected which disapear after the PCA correction: the feet go under the floor level and the arm movement is not natural (the subject walked with a healthy natural gate).
| Non-corrected gait motion | Corrected gait motion |
|---|---|
 |
|