Completed in the Summer of 2017, this project used used 6 IMU’s to estimate lower limb state. The wearable system was designed to support research in using functional electrical stimulation to correct gait abnormalities such as drop foot.
A peer-reviewed paper was written for Ingenium: Undergraduate Research Undergraduate Research at the Swanson School of Engineering. Citation available here.
A custom embedded system was made to interface with six MPU9250 IMU’s with a 1 khz update rate. The system included a Raspberry Pi, microcontroller, custom circuit boards, and substantial supporting software. Additionally a commerical Functional Electrical Stimulation Device was interfaced with the system. Sensor data could be forwarded through Wi-Fi in real-time for processing by simulink or a ROS application.
For a validation of the system a basic limb-angle estimation algorithm was designed using 6 instances of the popular Madgwick Filter. The following plot shows the results of a user taking a single (slow) step.