Abstract:
This work presents the process of dataset collection for the Human Activity Recognition
(HAR) including walking, running, stair climbing up and stair climbing down. The data
is collected via IMU sensors mounted on lower limbs and upper limbs. The dataset is
collected to train an algorithm for implementation on prosthetic knee joint to assist
amputees in auto calculation of gait pattern. A passive pneumatic knee joint is used to
semi automate it. This is done by designing a mechanism for controlling the damping of
pneumatic knee joint. The two damping screws i.e., flexion, extension, are located on the
rear of knee joint. In the scope of this project, they are controlled via selection switch
assembly.
