As part of my graduate thesis I worked on development of efficient inertial navigation algorithms under the guidance of Dr. Sangram Redkar. The central focus of the research was to enable usage of very low cost inertial sensors in accurate estimation human joint angles.
A wireless inertial measurement unit (IMU) was fabricated called as 'yIMU'. Cost of fabrication was under $15 and performance evaluation of the system has proven better accuracy in unaided yaw estimates compared to other similar units. This might allow usage of such low cost sensors in precision applications in pedestrian navigation in GPS denied areas.
. . .
Phase I: Fabrication of yIMU
Fabricated yIMU as strapdown system built on Arduino platform. The dual IMU system was based on the principle of common mode effect and exploited the redundancy of using an extra IMU to obtain superior raw inertial quantities. The sensor fusion algorithm was based on a computationally efficient complementary filter.
Phase II: Characterization experiments
Performed characterization of yIMU: Allan Variance analysis, temperature and vibration testing, to benchmark the performance of the system. Conducted orientation tracking experiments to ascertain the accuracy of the system.
Phase III: Joint Angle Tracking
Tested joint angle tracking performance of the system and obtained an average RMSE of 4.47 without external aiding of magnetometer. The system was tested against Noraxon inertial measurement system.
. . .
Deeply understood the concept of inertial tracking. Gained expertise in performing characterization of inertial sensors. Gained better understanding of systems engineering principles.
Robust Human Motion Tracking Using Low-Cost MEMS Inertial Sensors